|Home | About | Journals | Submit | Contact Us | Français|
Four homologous isoforms of glycerol-3-phosphate acyltransferase (GPAT), each the product of a separate gene, catalyze the synthesis of lysophosphatidic acid from glycerol-3-phosphate and long-chain acyl-CoA. This step initiates the synthesis of all the glycerolipids and evidence from gain-of-function and loss-of-function studies in mice and in cell culture strongly suggests that each isoform contributes to the synthesis of triacylglycerol. Much work remains to fully delineate the regulation of each GPAT isoform and its individual role in triacylglycerol synthesis.
After Eugene Kennedy and his colleagues showed that the esterification of glycerol-3- phosphate with a long-chain acyl-CoA was the initial step in the synthesis of phospholipids  and triacylglycerol (TAG) , Pullman’s group reported that this sn-glycerol-3-phosphate acyltransferase activity (GPAT; EC 220.127.116.11) was comprised of what appeared to be two isoforms, one located in the mitochondrial outer membrane and the other in the endoplasmic reticulum . The endoplasmic reticulum (microsomal) activity was inhibited by sulfhydryl reagents such as N-ethylmaleimide (NEM) and exhibited no preference for particular acyl-CoA species, whereas the mitochondrial activity was resistant to NEM inactivation and preferred to use saturated acyl-CoAs like 16:0-CoA and 18:0-CoA . With the identification of four genes encoding separate GPAT isoenzymes [5–11], we now know that GPAT mediated regulation of glycerolipid synthesis is more complex than anyone had previously thought; investigators are currently struggling with the question as to why four separate isoforms are required for glycerolipid biosynthesis.
Gpat1, the first mammalian GPAT isoform cloned [5, 6], resides in the outer mitochondrial membrane, is resistant NEM inactivation and prefers to use saturated acyl-CoAs . A second mitochondrial GPAT, GPAT2, also resides in the outer mitochondrial membrane, but its activity is inhibited by NEM and it has no long-chain acyl-CoA preference . The NEM-sensitive endoplasmic reticulum isoforms, GPAT3 and GPAT4, were identified very recently [9–11].
All four GPAT isoforms are members of the pfam 01553 family of glycerolipid acyltransferases and contain four conserved motifs first identified by a bioinformatics approach  (Figure 1). Mutagenesis of invariant residues in these motifs verified that these highly conserved regions are essential for the activity of E. coli GPAT (PlsB) [14, 15], mouse GPAT1 , and human dihydroxyacetone phosphate acyltransferase (DHAPAT) . Residues important for catalysis are the invariant histidine and aspartate in Motif I, the phenylalanine, glutamate, glycine, and arginine in Motif III, and the proline in Motif IV [15, 16, 18]. Amino acids important for binding sn-glycerol-3-phosphate are the phenylalanine and arginine in Motif II, and glutamate and serine in Motif III [15, 16, 18]. A diverse array of acyltransferases have been assigned to pfam 01553 and activities have been verified biochemically for GPAT1 - 4, sn-1-acylglycerol-3-phosphate O- acyltransferase (AGPAT) -1 and -2, DHAPAT, lysophosphatidylcholine acyltransferase (LPCAT) -1, -2 and -3, lysophosphatidylglycerol acyltransferase (LPGAT), and acyl-CoA:lysocardiolipin acyltransferase (ALCAT) [4, 17, 19–25]. In spite of its homology to the glycerolipid acyltransferases, tafazzin, a protein required for the synthesis of normal cardiolipin, exhibits a transacylase activity [26–28]. A fifth motif present in the AGPAT isoforms, GPAT-3 and -4, and tafazzin has been inferred because amino acid mutations in this region of human AGPAT2 and in tafazzin cause congenital generalized lipodystrophy or Barth syndrome, respectively [29, 30]. Amino acids important for the specialized activities of most of the other pfam 01553 family members have yet to be identified.
Although GPAT1 resides in the outer mitochondrial membrane and diacylglycerol acyltransferase, the terminal enzyme of TAG synthesis, resides on the ER, many studies support an important role for GPAT1 in regulating TAG synthesis. GPAT1 activity is highest in rat liver and adipose, tissues with a high capacity for TAG synthesis . In these tissues, GPAT1 activity is modulated in a manner consistent with the regulation of TAG synthesis. For example, in rat liver and adipose tissue, a 48 h fast decreases GPAT1 protein expression and activity more than 30%. When fasted rats are refed sucrose, GPAT1 protein expression and activity increases greater than 2-fold in liver [31, 32], and in perfused rat liver, insulin increases GPAT1 activity 34% . Similarly, streptozotocin-induced diabetes decreases GPAT1 activity in rat epididymal fat 60%, and insulin administration restores GPAT activity .
Changes in GPAT1 activity are mediated transcriptionally. When fasted mice are refed a high- carbohydrate diet, hepatic Gpat1 mRNA levels increase 20-fold due to enhanced transcription rates . The refeeding-induced increase in Gpat1 message occurs in streptozotocin-diabetic mice only if insulin is administered. This regulation of Gpat1 by insulin is mediated by sterol regulatory element binding protein-1 (SREBP-1), a transcription factor responsible for transactivating numerous genes required for enhanced fatty acid and TAG synthesis . Insulin induces transcription of SREBP-1c, and in a typical counter-regulatory fashion, glucagon via cAMP opposes the action of insulin . SREBP-1c appears to be the primary transcriptional regulator of GPAT1. The mouse Gpat1 promoter region, first characterized by Jerkins et al. , contains three SREBP-1 consensus sites which are responsible for SREBP-1 mediated transactivation . When SREBP-1c is ectopically expressed in 3T3-L1 preadipocytes, Gpat1 mRNA expression increases 6.7-fold . Conversely, in SREBP-1 knockout mice, the normal 8-fold induction of hepatic Gpat1 mRNA expression by refeeding is abolished . Although LXR agonists have been reported to upregulate GPAT1, LXR does not directly transactivate Gpat1, but rather upregulates SREBP-1c by a transcriptional mechanism . In addition, although Gpat1 mRNA increases with carbohydrate feeding, there is no evidence that this upregulation is mediated through the carbohydrate response element; instead, it appears to proceed entirely through insulin-mediated transactivation of SREBP-1c since Gpat1 mRNA expression is similar in wildtype and Chrebp−/− mice .
The studies summarized above indicate that GPAT1 in liver is regulated in a manner consistent with a role in initiating TAG synthesis when an animal is presented with excess carbohydrate or fat calories. Less information is available about the regulation of GPAT1 in other tissues. Gpat1 mRNA increases 10-fold when 3T3-L1 cells differentiate into adipocytes , coinciding with an increase in TAG synthesis, and mitochondrial GPAT activity (probably GPAT1) decreases 50% in rat epididymal fat and liver, but not in gastrocnemius muscle, after treadmill exercise, probably regulated by AMP-activated kinase (AMPK) . However, NEM-resistant GPAT activity increases in brown adipose tissue (BAT) when animals are exposed to cold , suggesting that TAG synthesis may continue simultaneously with TAG lipolysis. Such a futile cycle would contribute to the energy wasting and heat production in BAT. None of these effects, however, has been examined systematically and the controlling factors have not been elucidated.
Gain-of-function and loss-of-function studies highlight the importance of GPAT1 in de novo TAG synthesis. Plasmid- and adenovirus-mediated overexpression of GPAT1 in CHO and HEK293 cells and in primary rat hepatocytes increases TAG content and oleate incorporation into TAG 3- to 4-fold [44, 45]. Further, compared to rats infected with a control virus, when rats are infected with a GPAT1 expressing adenovirus, NEM-resistant GPAT activity, hepatic TAG content, and plasma TAG concentration each increase about 2.7-fold . In contrast, compared to wildtype controls, female mice that lack GPAT1 (Gpat1−/−) weigh less, have smaller gonadal fat pads, have 40% lower hepatic TAG content, 15% lower plasma TAG, and 30% reduced secretion of VLDL-TAG . Lipogenic diets also result in less TAG accumulation in liver and heart from Gpat1−/− mice [48, 49] (T.M. Lewin, unpublished), although the total weight gain of Gpat1−/− and wildtype mice is similar. Consistent with a role for GPAT1 in mediating TAG synthesis, a 90% adenovirus-mediated shRNA knockdown of liver Gpat1 in ob/ob mice reduces liver TAG content 30 –42% within 5 days . Surprisingly in this model, no change was observed in plasma TAG concentration.
In addition to controlling the flux of acyl-CoAs that enter the pathway of TAG synthesis, GPAT1 also regulates acyl-CoA use for β-oxidation. Once fatty acids are activated to acyl-CoAs, they can be esterified by GPAT and enter the pathway of glycerolipid biosynthesis or they can be converted to acyl-carnitines by carnitine palmitoyltransferase-1 (CPT1) and then enter the mitochondrion for β-oxidation. GPAT1 and CPT1 are intrinsic proteins of the outer mitochondrial membrane, and appear to compete for the same long-chain acyl-CoA pool in order to channel acyl-CoAs towards the synthetic or degradative pathways. This reciprocal channeling of acyl-CoAs is observed in Gpat1−/− livers which contain elevated acyl-CoAs and acyl-carnitines and show increased ketogenesis . Conversely, adenovirus mediated overexpression of GPAT1 in hepatocytes diminishes [14C]oleate oxidation 40% while increasing [14C]oleate incorporation into TAG more than 2-fold .
AMPK acutely partitions fatty acids to degradative vs. biosynthetic fates. AMPK is active when energy stores are deficient or when energy demand is high . High AMPK activity favors catabolic processes that supply ATP while inhibiting biosynthetic reactions that use ATP. Thus, active AMPK phosphorylates and inhibits acetyl CoA carboxylase, decreasing cellular malonyl- CoA, and relieving the malonyl-CoA inhibition of CPT1 . Because CPT1 is the rate-limiting step for β-oxidation, increased CPT1 activity results in enhanced fatty acid catabolism. Concomitantly, activated AMPK decreases GPAT1 activity and diminishes the use of acyl-CoAs for TAG synthesis. Specifically, both 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside, a nucleoside activator of AMPK, and recombinant AMPK inhibit GPAT activity 30–40% in hepatocytes and muscle . In hepatocytes, AICAR treatment diminishes [14C]oleate incorporation into TAG 40% . Thus, the actions of AMPK contribute to the acute reciprocal and coordinated regulation of acyl-CoA channeling at the outer mitochondrial membrane. Although GPAT1 contains several putative AMPK phosphorylation sites, it is not known whether AMPK directly phosphorylates GPAT1.
Casein kinase (CK2) phosphorylates GPAT1 and enhances GPAT1 activity 50–200% in liver  and in T-lymphocytes . Because CK2 also activates liver CPT1 20–25% and diminishes CPT1 inhibition by malonyl-CoA , in appears that unlike AMPK, CK2 does not reciprocally regulate fatty acid degradation and esterification. Further studies will be required to elucidate the physiological significance of CK2 phosphorylation of GPAT1 in the face of simultaneous activation of CPT1.
In addition to initiating TAG synthesis, GPAT1 also affects the synthesis of phospholipids. Consistent with the preference of GPAT1 to esterify palmitate at the sn-1 position of glycerol-3- phosphate, liver TAG and total phospholipids from Gpat1−/− mice contain 40% and 35% less 16:0, respectively, than lipids from control mice . In these mice the major phospholipids in liver, phosphatidylcholine (PC) and phosphatidylethanolamine (PE) contain 20% less 16:0 at the sn-1 position. In place of 16:0, PE contained 30% more 18:0 and 64% more 18:1. Changes in PC were similar, with 44% more 18:0 found at the sn-1 position. These changes in the amount of 16:0 at the sn-1 position resulted in an altered composition of fatty acids found at the sn-2 position. For PE from Gpat1−/− liver, the sn-2 position contained 36% more 20:4, 75% more 20:5, and 28% less 22:6 than wildtype mice. Similar results were found at the sn-2 position of PC which contained 40% more 20:4, 105% more 20:5, and 18% less 22:6. These data suggest that alterations in GPAT1 activity can alter both phospholipid fatty acid composition and cellular TAG content. Because PC is the primary component of the lipid monolayer surrounding lipid droplets, changes in PC fatty acid composition could alter lipid droplet structure.
GPAT2 was first recognized in liver mitochondria from Gpat1−/− mice because a polyclonal antibody raised against full-length GPAT1 identified a protein of similar molecular mass in liver from GPAT1 null mice . Characterization of GPAT2 activity in liver mitochondria from Gpat1−/− mice showed that, unlike GPAT1, GPAT2 is sensitive to inactivation by NEM, and has no preference for 16:0-CoA compared with oleoyl-CoA . Subsequent cloning of Gpat2, showed that the mRNA transcript was highly abundant in testis and relatively low in other tissues . When GPAT2 was overexpressed in Cos-7 cells, 84% more oleate was incorporated into TAG than in control transfected cells . Although studies in cultured cells indicate that GPAT2 can initiate TAG synthesis, GPAT2 does not substitute functionally for GPAT1. For example, despite being active in Gpat1−/− mice , the presence of GPAT2 does not prevent the diminished amount of TAG in liver  Consistent with this finding is that Gpat2 mRNA abundance is not altered by fasting or refeeding  as is observed for Gpat1 and other lipogenic enzymes like fatty acid synthase and acetyl-CoA carboxylase . The function of GPAT2 in testes mitochondria remains unexplored.
Until recently, it was believed that only a single NEM-sensitive (microsomal) GPAT existed and most studies earlier than 2007 could report only on enzyme activity. In both developing neonatal rat liver  and differentiating 3T3-L1 adipocytes , the microsomal activity increases 70-fold, but the microsomal GPAT activity seemed to be less influenced by hormonal and nutritional status than GPAT1. For example, activation of AMPK in rat hepatocytes decreases mitochondrial GPAT activity without affecting microsomal GPAT , and work by Saggerson’s lab in the 1970’s showed that the NEM-insensitive GPAT (presumably GPAT1) activity in rat liver is increased by insulin  and decreased by starvation , whereas the microsomal activity did not change. In adipose tissue however, other studies showed inactivation of microsomal GPAT activity by a tyrosine kinase and activation by insulin or sodium orthovanadate [62–65].
Now that two ER-located GPAT isoforms, GPAT3 and GPAT4, have been cloned, the specific regulation of these isoforms must be reinvestigated. It will be important to reevaluate specific regulation that may increase or decrease only a single isoform when TAG formation is stimulated or inhibited. Because the microsomal isoforms are differently expressed in rodent liver and in specific adipose depots, it is possible that changes in one of these isoforms or opposite changes in both of them resulted in little change in total GPAT activity, thereby misleading previous investigators.
The gene encoding Gpat3 was mislabeled in GenBank as AGPAT8 (LPAAT-theta), based on sequence homology to AGPAT1 and 2. Cao and colleagues identified GPAT3, based not only on sequence similarities to acyltransferases, but also by cross-referencing genes that increased during adipocyte differentiation , because microsomal GPAT activity had been reported to increase 70-fold during 3T3-L1 differentiation . Mouse and human Gpat3 encode 438- and 434- amino acid proteins, respectively, both with an approximate molecular mass of 50 kDa and exclusively expressed in the endoplasmic reticulum . In differentiated 3T3-L1 adipocytes, siRNA-mediated knockdown showed that GPAT3 (60% of mRNA) accounts for nearly 55% of total GPAT activity . Further, Gpat3 mRNA expression increases 60-fold in adipocytes after differentiation, and the Gpat3 transcript increases 4.5-fold in white adipose tissue from mice treated with rosiglitazone , consistent with the role of PPAR? in the regulation of lipogenic genes. Gpat3 is highly expressed in epididymal adipose tissue as well as small intestine in mice, and in kidney, testis, heart, skeletal muscle, and thyroid in humans [9, 66].
Gpat3 expression is relatively low in livers from both mice and humans , whereas NEM-sensitive, microsomal GPAT activity accounts for at least 50% of the total GPAT activity in liver . These data suggested that additional microsomal GPAT isoforms existed, and because GPAT3 shared 86.6% identical amino acid residues with a closely related protein, a fourth member of the GPAT family, GPAT4 was discovered . Like GPAT3, GPAT4 was misidentified as an AGPAT (AGPAT6 or LPAAT-zeta in GenBank) based on sequence homology. GPAT4, which is located exclusively in the endoplasmic reticulum, has 456 amino acids and a predicted molecular mass of 52.2 kDa, but is expressed as a 48 kDa protein, likely due to cleavage of a 38 amino acid signal peptide [67–69]. Like GPAT3 , GPAT4 can use a range of acyl-CoA substrates from 12 to 20 carbons, however the highest GPAT activity occurs with acyl-CoA species of 16- and 18-carbons, regardless of the level of saturation [10, 11]. Consistent with GPAT’s role in initiating TAG synthesis, Gpat4 mRNA is highly expressed in mouse lipogenic tissues, including liver and both BAT and visceral white adipose tissues, including omental, epididymal, and retroperitoneal depots . Gpat4 mRNA is also highly expressed in mouse testis .
In the liver, Gpat4−/− (formerly termed Agpat6−/−) mice have 49% lower total GPAT activity than wildtype mice and 65% less NEM-sensitive GPAT activity . In BAT, loss of Gpat4 accounted for 50% of total GPAT activity and 65% of NEM-sensitive GPAT activity. However, no changes in NEM-sensitive GPAT activity were observed in parametrial white adipose tissue . These results suggest that more than one NEM-sensitive GPAT functions in the liver and BAT and that different fat depots contain different complements of GPAT isoforms. Although Gpat3 is moderately expressed in the liver and BAT , its specific activity is unknown; therefore the relative contributions of GPAT3 and GPAT4 to NEM-sensitive GPAT activity in normal liver and BAT remain unclear. Importantly, no decrease in AGPAT activity was detected in tissues from the Gpat4−/− mice, solidifying the discovery of GPAT4 as the enzyme previously identified as AGPAT6 .
Suggesting an important role for GPAT4 in TAG synthesis, Gpat4−/− mice have a 45% lower content of TAG in liver and, when fed a chow diet, have a significantly decreased TAG content of both BAT and gonadal white adipose depots . The reduction of adipose tissue contributes to the ~25% decrease in body weight but does not explain it fully. Further, Gpat4−/− mice are protected from both high fat/high carbohydrate diet-induced obesity and from genetic obesity when they are crossed with obese, leptin deficient ob/ob mice . Subdermal adipose tissue is greatly reduced in Gpat4−/−mice. It remains to be determined whether the reductions in adipose tissue are due to attenuated TAG production or to inhibited adipocyte differentiation as occurs with congenital generalized lipodystrophy due to mutations in AGPAT2 . Despite increased energy expenditure and reduced adipose mass and hepatic TAG content, Gpat4−/− mice have no significant differences in fasting plasma glucose or insulin concentrations and can maintain their body temperature when housed at 4 °C for 4 h . Gpat4 mRNA is also highly expressed in mammary epithelium, where lipids are synthesized for milk production. The content of lipid droplets in mammary epithelial cells is significantly reduced in Gpat4−/− mice, suggesting that GPAT4 has an integral role in milk lipid synthesis . Because the content of diacylglycerol and TAG in milk is reduced approximately 90%, most pups born to Gpat4−/− mice die within the first 48 hours after birth, unless nursed by a foster mother.
Overexpression of GPAT4 does not increase TAG production in either HEK293 cells  or Cos-7 cells, despite a 60% increase in NEM-sensitive GPAT activity . Because livers from Gpat4−/− mice contain 45% less TAG than livers from wildtype mice, we assume that GPAT4 in vivo normally initiates the synthesis of TAG  and that the cultured cells lacked some cellular machinery required to enable the esterification of the excess DAG synthesized under these conditions . However, DAG synthesized via GPAT1 in the same cultured cells was able to increase incorporation of [ 14C]oleic acid into both diacylglycerol and TAG pools [10, 44]. Overexpression of GPAT4 in cultured cells increased [14C]oleic acid incorporation into phosphatidylinositol , but overexpression of the highly homologous GPAT3 in HEK293 cells increased [14C]oleic acid incorporation only into TAG and not phospholipids . These data suggest that the individual GPAT isoforms may generate distinct lipid intermediates destined for specific biosynthetic pathways. .
We are presently faced with a plethora of GPAT isoforms; all four isoforms are apparently able to initiate the glycerolipid synthetic pathway and at least three of these isoforms are able to increase the incorporation of fatty acid into TAG. Although overexpression of GPAT4 in cultured cells curiously did not increase TAG synthesis, the diminished hepatic TAG content in Gpat4−/− mice suggests that its LPA product, too, can be incorporated into TAG.
GPAT has always been considered to be the rate-limiting step in the pathway of glycerolipid synthesis and to regulate fatty acid flux through the pathway (Figure 2). It seems probable that the biosynthesis of the two major lipid components of lipid droplets, the phospholipid monolayer and the core TAG, must be synchronized. Thus, it is likely that the different GPAT isoforms work in a coordinated way to synthesize these lipids. In that sense, it is puzzling that a marked increase in TAG synthesis is not accompanied by enhanced synthesis of phospholipids, primarily phosphatidylcholine, required to coat the surfaces of newly forming lipid droplets . Further, growing evidence indicates that the lipid intermediates in the glycerolipid synthetic pathway may initiate intracellular signaling via protein kinase C and similar pathways [46, 49]. It is likely that the regulation of lipid droplet associated proteins is also inherent to lipid synthesis. Much work remains to fully delineate the transcriptional and acute regulation of each GPAT isoform as well as its individual role in the synthesis of TAG, specific phospholipids, and lipid intermediates involved in signaling.
Supported by NIH DK56598 (RAC) and a UNC CNRU (DK056350) Young Investigator Award to TML.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.