The Enzymes of Neutral Lipid Synthesis*

Fueled by fat-rich diets and sedentary lifestyles, atherosclerosis and obesity are major health issues in the Western world. Both diseases involve excessive accumulation of neutral lipids: cholesterol esters in arteries in atherosclerosis and triglycerides in adipocytes in obesity. The study of these diseases has sparked interest in the biochemistry and molecular biology of neutral lipid synthesis.

involving trans-splicing of two discontinuous precursor RNAs produced from chromosomes 1 and 7 (21). The origin of the 7.0-kb transcript is unknown. The human ACAT2 gene encodes a single mRNA of 2.2 kb, and the human DGAT gene encodes two mRNAs of 2.0 and 2.4 kb. Regions of shared sequence identity are present throughout the ACAT and DGAT proteins, but the greatest similarity is found in their C termini (Fig. 2). Human and mouse ACAT2 are ϳ40% identical to human ACAT1, and DGAT is ϳ20% identical to ACAT1.
Little is known about the relationship between protein structure and function in ACAT/DGAT family members. One motif, FYXD-WWN (amino acids 403-409 of human ACAT1), is highly conserved in all family members and may be involved in fatty acyl-CoA binding. Another motif, MKXXSF (amino acids 265-270 of human ACAT1), is conserved in ACAT family members. The serine of this motif is required for ACAT activity (22,23). All three family members have a potential tyrosine phosphorylation motif and at least one N-linked glycosylation site, both of unknown significance. ACAT1 and ACAT2 contain an N-terminal leucine zipper motif. Although ACAT1 functions as a homotetramer (24), it is unknown whether this leucine zipper motif participates in tetramer formation or whether ACAT2 and DGAT act as multimers. ACAT1 and ACAT2 do not form hetero-oligomeric complexes (25).
All ACAT/DGAT proteins possess multiple hydrophobic regions predicted to serve as transmembrane domains, although there is not yet a consensus structural model for the proteins. Investigations of human ACAT1 topology have yielded different results. One study, performed by inserting epitopes at different positions in the expressed protein (26), found seven transmembrane domains. Another study, performed by expressing truncated forms of the protein (22), found five transmembrane domains. In both studies, the N terminus of the protein localized to the cytoplasm and the C terminus to the ER lumen (22,26). Topologic studies of human ACAT2 performed with the truncation mutant method also found five transmembrane domains with the N terminus in the cytoplasm and the C terminus in the ER lumen (22). The latter study found that a serine residue essential for ACAT activity was located in the cytoplasm for ACAT1 and in the ER lumen for ACAT2, suggesting that the active sites of these enzymes reside on opposite sides of the ER membrane. Several putative transmembrane sequences are conserved in ACAT1 and ACAT2 but not in DGAT (26), suggesting that these regions are involved in cholesterol binding. The topology of DGAT has not been reported, but a putative diacylglycerol binding motif has been identified (27).

Tissue and Cellular Distributions
The tissue distributions of ACAT1 and ACAT2 are largely complementary. ACAT1 mRNA is present in many tissues, with the highest levels in macrophages and in adrenal and sebaceous glands (12,28,29), all of which store cholesterol esters in cytoplasmic droplets. ACAT1 is also highly expressed in human atherosclerotic lesions, particularly in macrophage foam cells (30). ACAT2 is expressed predominantly in the liver and small intestine (27,31,32). In humans, nonhuman primates, and mice, ACAT2 appears to be the major ACAT in the small intestine (25,33,34) and the predominant isozyme in hepatocytes of nonhuman primates and mice (33,34).
Whether ACAT1 or ACAT2 is the predominant enzyme in human hepatocytes is controversial. Both are expressed in human hepatocytes and HepG2 cells (25). However, immunodepletion experiments found that ACAT1 accounts for ϳ90% of ACAT activity in human liver (35), whereas ACAT2 accounts for most of the ACAT activity in adult small intestine and fetal liver but only 10 -20% of activity in adult liver (25). It is unclear whether the hepatic ACAT1 activity resides in hepatocytes or macrophage-derived Kupffer cells. ACAT1 expression was detected immunohistochemically in both cell types, with stronger staining in Kupffer cells (36 nonhuman primates, ACAT1 was expressed only in Kupffer cells in the liver (33). A significant proportion of the ACAT1 activity in human liver thus may be attributable to Kupffer cells. Another study found little evidence of ACAT2 expression in human hepatocytes (25). DGAT mRNA and activity are ubiquitous in mouse and human tissues, with the highest levels in liver, small intestine, and adipose tissue (11,27,37,38) and somewhat lower levels in testis and adrenal gland (11). This wide range of tissue expression is consistent with the involvement of DGAT in the glycerol phosphate pathway of triglyceride synthesis, which is common to most cells. DGAT protein expression in tissues has not been studied because of the lack of suitable antibodies.
The intracellular localization of enzymes in this class has been examined only for ACAT1. In human melanoma cells and fibroblasts and in mouse macrophages, ACAT1 is located primarily in the ER (36,39,40). In mouse macrophages, a small portion of ACAT1 immunoreactivity localizes to a region near the trans-Golgi network (40,41). In macrophages, ACAT1 localization may change with different conditions (36,40).

Biochemical Activity and Regulation
ACAT and DGAT substrates have been identified by expressing the proteins in insect cells or yeast devoid of sterol esterification activity. In insect cells, mouse and human ACAT1 and ACAT2 utilize a variety of oxysterols in addition to cholesterol as substrates (31,42). Ergosterol and plant sterols are poor substrates (31,43). Both enzymes also utilize a wide variety of long chain fatty acyl-CoAs as substrates (31), although arachidonyl-CoA appears to be a better substrate for ACAT1 than ACAT2 (31,43). In insect cells, DGAT utilizes only diacylglycerol as the acyl acceptor (37). DGAT has a broad fatty acyl-CoA specificity (37). ACAT1 is regulated primarily by post-translational mechanisms (3). In cholesterol-loaded cells, esterification increases without changes in the expression level of ACAT1 mRNA (19,20,44) or protein (20,24,45). Purified ACAT1 is activated allosterically by cholesterol or oxysterols (46). ACAT1 activity may also be regulated by its intracellular localization. For example, a portion of ACAT1 immunoreactivity in macrophages is found near the trans-Golgi network and the endocytic recycling compartment (41). These cholesterol-rich organelles are involved in membrane traffic with the plasma membrane, and the proximity of ACAT1 to these organelles may be important for cholesterol ester synthesis during the formation of macrophage foam cells. The observation that 30 -40% of ACAT1 redistributed from the ER to small vesicles during foam cell formation is consistent with this possibility (36). Regulation by phosphorylation has not been demonstrated.
Little is known about the regulation of ACAT2 and DGAT. Levels of DGAT mRNA increase markedly in parallel with enzyme activity in NIH 3T3-L1 cells during their differentiation into adipocytes (37). This finding suggests that transcription factors involved in adipogenesis activate DGAT expression. Like other family members, DGAT possesses a putative tyrosine phosphorylation site. The use of this site in DGAT regulation has not been demonstrated, although two reports (49,50) have indicated that the enzyme may be post-translationally regulated by a tyrosine kinase.

Functional Analysis of ACAT1 and ACAT2
Gene knockout experiments in mice have provided insights into the biological roles of ACAT1 and ACAT2. In ACAT1-deficient (ACAT1 Ϫ/Ϫ ) mice, plasma cholesterol levels, hepatic cholesterol esterification activity, and intestinal cholesterol absorption are normal (51), consistent with the presence of a second ACAT enzyme in the small intestine and liver of mice. ACAT1 deficiency has its greatest effect in the adrenal glands and macrophages. Cholesterol esters and cholesterol esterification activity are virtually undetectable in the adrenal cortex of ACAT1 Ϫ/Ϫ mice, although the depletion of cholesterol esters does not affect corticosterone synthesis (51). In addition, peritoneal macrophages from ACAT1 Ϫ/Ϫ mice do not accumulate cholesterol esters when incubated with cholesterolcontaining lipoproteins (51). This finding provides a unique tool for testing the role of ACAT1 in macrophage foam cell formation and atherosclerosis (see below). Another ACAT1 gene disruption study in mice confirmed general aspects of the ACAT1 Ϫ/Ϫ phenotype and showed that ACAT1 deficiency reduces cholesterol esterification activity by Ͼ95% in testis and ovary and results in atrophy of the meibomian glands of the eyelids (52). A naturally occurring mutation in the mouse ACAT1 gene is present in the AKR strain of inbred mice (53), which are homozygous for the adrenocortical lipid depletion (ald) allele (54).
The analysis of ACAT2-deficient mice (ACAT2 Ϫ/Ϫ ) (34) complemented findings from ACAT1 Ϫ/Ϫ mice. ACAT2 Ϫ/Ϫ mice have nearly undetectable ACAT activity and lack cholesterol esters in the liver and small intestine (34). Because they cannot synthesize cholesterol esters in the small intestine, ACAT2 Ϫ/Ϫ mice have a reduced capacity to obtain cholesterol from the diet. As a result, they do not develop hypercholesterolemia or form cholesterol gallstones when fed a diet high in fat, cholesterol, and cholic acid (34). ACAT2 deficiency markedly decreases the cholesterol ester content of apolipoprotein (apo) B-containing lipoproteins in mice fed either a chow diet or the above diet (34). Thus, ACAT2 is the cholesterol esterification enzyme in mouse liver and small intestine and plays a major role in mediating the response to dietary cholesterol.
Gene disruption studies suggest that both ACAT enzymes synthesize cholesterol esters: ACAT1 for storage in cytosolic droplets and ACAT2 for secretion in apoB-containing lipoproteins. However, these results may reflect their different tissue distributions rather than inherent biochemical differences. Both ACAT1 and ACAT2, when expressed in a cell line deficient in cholesterol esterification activity (AC29 cells), synthesize cholesterol esters that are stored in cytoplasmic droplets (25). Also, overexpression of ACAT1 in mouse liver causes both cholesterol ester storage and secretion (55).

Role of ACAT1 in Atherosclerosis
Because cholesterol ester synthesis by ACAT1 is involved in macrophage foam cell formation, it has been hypothesized that inhibition of ACAT1 may reduce atherosclerosis. To test the hypothesis that ACAT1 deficiency in macrophages inhibits atherosclerosis, ACAT1 Ϫ/Ϫ mice were crossed with two strains of mice that are susceptible to atherosclerosis: low density lipoprotein receptor-deficient (LDLR Ϫ/Ϫ ) mice and apoE-deficient (apoE Ϫ/Ϫ ) mice (56). ACAT1 deficiency did not prevent atherosclerotic lesions in either model. Because ACAT1 deficiency also reduced serum cholesterol levels in both strains of mice, its effect on lesion size was

Minireview: Enzymes of Neutral Lipid Synthesis 40370
confounded by effects on serum cholesterol levels, and thus the contribution of ACAT1 to plaque formation could not be rigorously determined. Qualitative analysis of lesions in the mutant mice, however, revealed reduced amounts of cholesterol esters and macrophages (56). These alterations may increase lesion stability (57), although this hypothesis remains to be tested.
Studies that analyzed the quantitative effect of ACAT1 deficiency on lesion development have produced different results. Ishibashi and colleagues (52) crossed a different strain of ACAT1 Ϫ/Ϫ mice with LDLR Ϫ/Ϫ and apoE Ϫ/Ϫ mice and found that ACAT1 deficiency was associated with ϳ50% reduction in aortic root lesions in LDLR Ϫ/Ϫ mice and did not decrease serum cholesterol levels (52). However, Fazio et al. (58) showed that LDLR Ϫ/Ϫ mice reconstituted with ACAT1 Ϫ/Ϫ macrophages had increased atherosclerotic lesions that contained more free cholesterol and fewer macrophages than control lesions. The apparent discrepancies between these studies may reflect differences in methodology or study design. Ishibashi and colleagues (52) quantified lesion areas by measuring neutral lipid staining even though the cholesterol ester content of atherosclerotic lesions in ACAT1 Ϫ/Ϫ mice was significantly diminished (56,58). The lesion areas consequently may have been underestimated. In the study by Fazio et al. (58), selective inhibition of macrophage ACAT1 may not have reduced lesion size because ACAT activity was not inhibited in adjacent arterial smooth muscle cells, which can also become foam cells.
Both complete (52,56) and macrophage-specific (56) ACAT1 deficiencies have deleterious effects in hypercholesterolemic mice. ACAT1 deficiency causes extensive deposition of free cholesterol in the skin (52, 56) and brain (56) of LDLR Ϫ/Ϫ and apoE Ϫ/Ϫ mice, suggesting an important role for the enzyme in cholesterol metabolism in these tissues when serum sterol levels are elevated. Although the dramatic effects in these models reflect the extreme condition of complete ACAT1 inhibition in the face of severe hypercholesterolemia, they indicate that selective ACAT1 inhibition can be detrimental.

Functional Analysis of DGAT
Because DGAT catalyzes the final step in the major pathway of triglyceride synthesis (8,10), it was speculated that inactivation of the encoding gene in the mouse (Dgat) would cause severe morbidity or early lethality. Surprisingly, DGAT-deficient mice are viable and appear healthy although adult mice develop dry fur and female mice have a lactation defect (38). Even though these animals lack triglyceride synthesis in most tissues examined, Dgat Ϫ/Ϫ mice have normal serum triglyceride levels and triglycerides in white adipose tissue (38). These results indicate the existence of alternative pathways of triglyceride synthesis (11). 2 Although DGAT deficiency does not prevent triglyceride synthesis, Dgat Ϫ/Ϫ mice have ϳ50% reduction in fat pad content and are resistant to weight gain when fed a high fat diet (38). Unexpectedly, these findings are associated with an ϳ15% increase in daily total energy expenditure, part of which is mediated by increased physical activity when the mice are fed a high fat diet (38). The mechanism for the increased energy expenditure is unknown, but the Dgat Ϫ/Ϫ phenotype indicates that the disruption of triglyceride synthesis significantly affects energy metabolism.

ACAT and DGAT Enzymes as Pharmaceutical Targets
Because ACAT enzymes participate in cholesterol ester synthesis for assembly of atherogenic apoB-containing lipoproteins (59,60) and in macrophage foam cell formation, ACAT inhibitors have been developed as potential therapeutics (61)(62)(63)(64). These agents produced mixed results in several animal models. In cholesterol-fed rabbits and hamsters, ACAT inhibitors reduced atherosclerotic lesion development independently of their effects on serum cholesterol levels (57,(65)(66)(67). A study of genetic models of hypercholesterolemia in mice and rabbits suggested that ACAT inhibitors either had no effect or increased aortic sinus lesion size (68). One inhibitor (avasimibe) reduced atherosclerosis in several animal models (57,69) and is being evaluated in clinical trials. This compound, like most currently available ACAT inhibitors, is not selective for one of the two ACAT enzymes (31,69). This pan-specificity may be desirable because pharmacologic (68) and genetic studies (56,58) in animals suggest that specific inhibition of macrophage ACAT1 may increase lesion size, possibly because of toxicity from  2. Alignment of human ACAT1, ACAT2, and DGAT proteins. Conserved residues are shown in color. Asterisks indicate single, fully conserved residues; colons indicate strongly conserved residues; and periods indicate weakly conserved residues.
Minireview: Enzymes of Neutral Lipid Synthesis 40371 excess free cholesterol. Selective inactivation of ACAT1 in hypercholesterolemia may also have detrimental consequences (see above). Therefore, selective ACAT1 inhibition should be approached cautiously in humans (70). Partial inhibition of both ACAT enzymes may be a preferable strategy (71). A selective inhibitor of ACAT2 may be useful for preventing diet-induced hypercholesterolemia (34,72). Recent findings in Dgat Ϫ/Ϫ mice (38) have sparked interest in DGAT inhibitors as possible therapeutic agents for obesity. Several naturally occurring compounds have been identified (73)(74)(75).