Biochemical Journal

Review article

Hepatic triacylglycerol synthesis and secretion: DGAT2 as the link between glycaemia and triglyceridaemia

Victor A. Zammit

Abstract

The liver regulates both glycaemia and triglyceridaemia. Hyperglycaemia and hypertriglyceridaemia are both characteristic of (pre)diabetes. Recent observations on the specialised role of DGAT2 (diacylglycerol acyltransferase 2) in catalysing the de novo synthesis of triacylglycerols from newly synthesized fatty acids and nascent diacylglycerols identifies this enzyme as the link between the two. This places DGAT2 at the centre of carbohydrate-induced hypertriglyceridaemia and hepatic steatosis. This function is complemented, but not substituted for, by the ability of DGAT1 to rescue partial glycerides from complete hydrolysis. In peripheral tissues not normally considered to be lipogenic, synthesis of triacylgycerols may largely bypass DGAT2 except in hyperglycaemic/hyperinsulinaemic conditions, when induction of de novo fatty acid synthesis in these tissues may contribute towards increased triacylglycerol secretion (intestine) or insulin resistance (adipose tissue, and cardiac and skeletal muscle).

  • de novo fatty acid synthesis
  • diacylglycerol acyltransferase
  • heart
  • hypertriglyceridaemia
  • intestine
  • lipogenesis
  • liver
  • muscle

INTRODUCTION

TAGs (triacylglycerols; also known as triglycerides) perform multiple functions in mammalian cells and in the traffic of substrates between tissues. They are a major source of dietary energy, and the intricate mechanisms that have been developed to enable their digestion, uptake, distribution and storage attest to their importance as the ultimate high-capacity form of energy storage in vertebrates. White adipose tissue is specialized for their storage, and it is therefore not surprising that it should be the origin of multiple signals (adipokines) that feedback to the brain the longer-term energy status of the organism, and also act directly on other tissues. However, lipid droplets (of which TAG is a major component) exist in most cell types where they reflect the difference between the rate at which fatty acids reach tissues, or are synthesized therein, and the rate at which they are metabolized. This may occur through oxidation (e.g. in muscle), or, as in the liver and intestine, through packaging within TRLs (TAG-rich lipoproteins) which are secreted. TAG that is accumulated in tissues other than adipose is referred to as ‘ectopic’ and can be used as a local readily available source of energy for oxidative tissues (e.g. cardiac muscle, type-1 skeletal muscle fibres). But expansion of ectopic lipid levels tends to be associated with metabolic dysregulation and, in particular, altered glucose–fatty acid interactions. Thus hepatic steatosis results in the deterioration of liver function, the possibility of hepatic insulin resistance, and the development of inflammation, non-alcoholic steatohepatitis and associated morbidities [1]. Increased cardiac intramyocytic TAG is associated with cardiomyopathy and heart failure [25]. Elevated IMTG (intramyocellular TAG) in skeletal muscle, particularly in type-2 fibres, is associated with muscle insulin resistance which, owing to the large mass of skeletal muscle, is the greatest contributor towards whole-body insulin resistance [69]. The term ‘lipotoxicity’ has been applied to this phenomenon, not because TAG itself is deleterious, but because the intermediates of TAG synthesis and/or lipolysis are highly biologically active. For example, long-chain acyl-CoA are membrane-active, and can affect gene expression and ion channel activity (see [10]). Together with DAGs (diacylgycerols), they have been suggested to cause insulin resistance by activating PKC (protein kinase C) isoforms whose serine phosphorylation of insulin signalling cascade components results in the attenuation of insulin signalling [1115]. Similarly, intracellular levels of ceramide, which can be synthesized de novo from palmitoyl-CoA, are also known to correlate with increased insulin resistance in muscle [16]. Fatty acids and ceramides also both feed into the synthesis of pro-inflammatory pathways [e.g. NF-κB (nuclear factor κB)] and the synthesis of pro-insulin resistance factors [e.g. TNFα (tumour necrosis factor α)]. Therefore the extent to which ectopic TAG content in tissues is associated with an excessive supply of fatty acids tends to be a feature of metabolic pathological conditions, with one notable exception (see below). Lipotoxicity may not be the only mechanism that is involved in the aetiology of insulin resistance, but multiple other mechanisms {e.g. generation of ROS (reactive oxygen species) and ER (endoplasmic reticulum) stress [17]} may be related to, or overlap with, it [15].

TAG SYNTHESIS AND SECRETION

Most TAG formation can occur in one of two ways. First, through re-esterification of partial glycerides [DAG and MAG (monoacylglycerol)] that themselves arise from partial hydrolysis of pre-existing TAGs. In this instance, no net genuinely new synthesis of glyceride occurs. Secondly through the de novo incorporation of glycerol 3-phosphate into the glyceride entity by sequential esterification of its hydroxy groups, followed by the formation of DAG, and its final esterification to TAG. This involves genuine de novo glyceride synthesis. Hepatocytes have the capacity to do both. Together with enterocytes, they are specialized for the secretion of TRLs, VLDLs [very-LDLs (low-density lipoproteins)] and chylomicrons respectively. However, in enterocytes, most TAG formation represents a reassembly of the TAG molecule from the products of dietary TAG digestion by pancreatic lipase (i.e. 2-MAG and fatty acids). In the liver, TAG synthesis is an integral part of the mechanisms through which the liver orchestrates lipid metabolism for the whole body, whereby fatty acids mobilized from adipose tissue, or released into the circulation during lipoprotein lipase action on TRLs, are re-packaged into VLDLs and recirculated to the periphery.

However, the liver is also the tissue that maintains glucose homoeostasis, with glucose uptake and release modulated continually to maintain the plasma glucose concentration within relatively narrow limits. Therefore the liver would be expected to have the function of integrating glycaemia with triglyceridaemia, and to have mechanisms that link the availability of glucose and other saccharides (e.g. fructose) with its output of VLDL–TAG, which largely determines triglyceridaemia in obesity and Type 2 diabetes [1820]. It is of note that high-carbohydrate diets in humans and animal models are associated with hypertriglyceridaemia, adipose weight gain and their associated metabolic pathologies [2123]. This ability of the liver to link carbohydrate with TAG metabolism may account for the marked lack of success of the ‘low-fat’ dietary advice to diminish the burden of dyslipidaemia and associated cardiometabolic disease in industrialized populations over the last several decades.

SECRETION OF TRLs

In hepatocytes and enterocytes, the secretion of TRLs occurs through the assembly of particles containing a hydrophobic core of TAG and cholesteryl esters, within a hydrophilic coat of phospholipids, cholesterol and associated apoproteins. ApoB (apolipoprotein B) is the molecule that keeps the particle together by virtue of its repeated highly hydrophobic sequences. The mechanism of this assembly is similar in the two cell types, although that in hepatocytes has been much more extensively studied [24,25]. At the outset it needs to be emphasized that, although terms including ‘cytosolic’ and ‘luminal’ are used routinely, the hydrophobic nature of TAG necessitates that its synthesis and intracellular compartmentation are highly membrane-associated/dependent, although TAG metabolism still appears to distinguish on which aspect of the membrane it occurs or to which compartment the resulting TAG is destined for. Central to the assembly of nascent pre-VLDL particles is the co-translational insertion of the apoB molecule through the ERM (ER membrane). During this process, apoB abstracts TAG from within an inter-leaflet location of the ERM where it is synthesized into the lumen of the ER. There it surrounds this relatively small quantity of TAG to form lipid-poor nascent LDL-type particles which can be observed in the ER lumen [24]. This process requires the involvement of the MTP (microsomal transfer protein). It is the first step in the formation of small VLDL particles (VLDL2) [24]. However, this accounts for only approximately 30% of the TAG synthesized on the ERM. Most of the TAG is enveloped by the cytosolic leaflet of the ERM to form ‘cytosolic’ lipid droplets which are bounded by a single phospholipid layer, the proteome of which consists, although not exclusively, of many ER membrane proteins, in addition to others that are acquired independently and reversibly [2628].

There is a third morphologically and compartmentally distinct pool of TAG within hepatocytes that resides within the lumen of the smooth ER. Although having other proteins associated with them, these lipid droplets are devoid of apoB [29,30]. It is an important pool of TAG because it serves to enlarge or ‘mature’ the nascent lipoprotein particles into large VLDL, as the two meet within the proximal Golgi to undergo the second lipidation step; this step also appears to be MTP-dependent [31]. The extent to which this second-step lipidation occurs is presumed to determine the size distribution of the VLDL secreted (larger VLDL1 compared with smaller VLDL2). Larger VLDLs (which are more prevalent in diabetes [18,19]) determine the atherogenicity of the LDL that is formed after their TAG is hydrolysed by lipoprotein lipase in peripheral tissues and undergoes lipid exchange with HDL (high-density lipoprotein) in the circulation [19,32]. Therefore the ability to effect the second-step lipidation of nascent VLDLs through the transfer of TAG from ER luminal LDs (lipid droplets) to nascent lipoprotein particles is a process with important (patho)physiological consequences, and which may be amenable to pharmacological manipulation if we can understand the mechanism(s) through which this apoB-free pool of TAG originates and is utilized for VLDL enlargement and maturation.

The origin of this intra-ER luminal pool of TAG is thought to be the TAG within the cytosolic LDs, but only after hydrolysis followed by resynthesis [33]. Communication between TAG in the cytosolic lipid droplets and the ER luminal lipid droplets is, therefore, an important process which accounts for approximately 70% of overall TAG secretion by the liver. However, hydrolysis re-esterification occurs at a much higher rate than the rate of secretion (approximately 20-fold faster) [34]. The observations that TAG hydrolysis does not proceed all the way to glycerol and fatty acids, but to partial glycerides (DAG and MAG) [3436], gave rise to the concept [37,38] that the communication between the cytosolic and luminal pools of TAG is performed by DAG which, unlike TAG, is highly permeable through the lipid bilayer [39]. The corollary of this concept was that there would be DGAT (DAG acyltransferase) activity expressed on both sides of the ERM (i.e. across the barrier represented by the impermeability of the lipid bilayer to acyl-CoA) to enable TAG synthesis to occur on both aspects of the membrane.

Early observations showed that, indeed, DAG esterifying activity (DGAT activity) is expressed on both aspects of the membrane, i.e. that there are ‘overt’ and ‘latent’ DGAT activities of DGAT [37,38]. It was subsequently shown that acyl-CoA availability for TAG synthesis on the luminal aspect of the membrane may be completed by the transfer of acyl moieties as acylcarnitine esters [40,41]. Although the enzyme mechanism for acylcarnitine formation and transfer across the ER membrane is not as well studied as for the mitochondria, there are long-standing observations which suggest that an analogous system to that in mitochondria exists in the ERM [42]. This model accounted for the transfer of the glyceride entity across the ERM, while keeping separate the TAG in the cytosolic and ER luminal LDs [43]. The racemization of the stereoisomers of TAG between that in the liver and that of secreted VLDLs indicated that lipolysis–re-esterifiaction cycling occurred to DAG and MAG [35,36], but it did not exclude the possibility that cycling occurs also within the ER lumen, and that DAG moves in both directions and equilibrates across the ERM (see Figure 1).

Figure 1 DGAT2 and DGAT1 act in series

Pathway showing how DGAT2 acts upstream of DGAT1, and the relationship between de novo fatty acid synthesis, TAG synthesis and secretion in hepatocytes. DGAT2 acts upstream of DGAT1 and utilizes nascent DAG and de-novo-synthesized fatty acids as substrates. TAG formed by DGAT2 provides substrate for lipases which generate partial glycerides that, together with exogenous fatty acids, are substrates for DGAT1. DGAT2 utilises de-novo-synthesized (and desaturated) fatty acids preferentially, possibly by virtue of its co-localization with enzymes of the glycerol 3-phosphate pathway and SCD1 (see the text). The association with FATP1 facilitates deposition of newly synthesized TAG into small cytosolic LDs. A minority of the TAG formed is partly sequestered for secretion during the co-translational insertion of the apoB molecule through the ERM, and partly stored in cytosolic LDs, without undergoing lipolysis. Most of the newly DGAT2-synthesized TAG is hydrolysed by lipase(s) located in the cytosolic compartment of the cell (or structures exposed to it) to give partial glycerides that are re-esterified to TAG by DGAT1, using preformed exogenously derived fatty acids. DAG equilibrates across the ERM. DGAT1 activity expressed on the luminal aspect of the ERM uses DAG (derived from TAG lipolysis either in the cytosol or from TAG–DAG cycling within the ER lumen) to synthesize TAG within the ER lumen. This luminal (apoB-free) TAG is used for the second-step lipidation of nascent VLDL, before the mature VLDL particle is secreted. LP, lipoprotein. Modified from Wurie, H.R., Buckett, L. and Zammit, V.A. (2012) Diacylglycerol acyltransferase 2 acts upstream of diacylglycerol acyltransferase 1 and utilizes nascent diglycerides and de novo synthesized fatty acids in HepG2 cells. FEBS J. 279, 3033–3047, with permission.

TWO DISTINCT DGATs

The above observations were followed by the cloning of the cDNA of two unrelated proteins that have DGAT activity, which were termed DGAT1 and DGAT2, in the order in which they were described [4446]. A third enzyme present in human enterocytes {MGAT (MAG acyltransferase) 3 [47]} also has substantial DGAT activity, but will not be considered further in the present review. In mammals, DGAT1 is expressed in skeletal muscle, skin, intestine (ileum, colon) and testis, with lower levels of expression in liver and adipose tissue [44]. DGAT2 is similarly widely expressed with high expression levels in hepatocytes and adipocytes [45]. Both of the different tissue distributions, and the fact that the two major proteins with DGAT activity have been evolved from two separate protein families [48], indicate that these enzymes have specialized functions even though they catalyse the same reaction and are co-expressed, to different extents, in individual cell types. In addition, their respective overexpression in hepatoma-derived MzcArdle cells [49] results in markedly different intracellular patterns of TAG droplet accumulation within the cells, again suggesting that the properties of TAG synthesized by DGAT1 and DGAT2 may be distinct.

The salient physiological feature of DGAT1 and DGAT2 is that although they catalyse the same reaction, and are co-expressed in most cells, they are non-redundant, i.e. when the expression of one is disrupted, the other cannot substitute for it (although it has been claimed that DGAT1 can substitute for the absence of DGAT2 [49,50], those experiments only measured fatty acid incorporation into TAG, which can proceed without net new synthesis of TAG, through hydrolysis–re-esterification cycling, see above). This phenomenon is most apparent in the very different phenotypes of the respective global knockout mice (Dgat1−/− and Dgat2−/−). Dgat1−/− mice have a very mild phenotype and are resistant to obesity, and the development of diet-induced obesity and insulin resistance [51]. They have slightly lower hepatic TAG levels when fed a low-fat diet, but, importantly (see below) when maintained on a high-fat diet they have a significantly lowered hepatic TAG content [52,53]. In contrast Dgat2−/− mice do not survive for more that 24 h after birth and are severely lipopaenic, with water-barrier deficiencies in their skin [49]. The inability of DGAT1 (the expression of which is not increased) to rescue the Dgat2−/− phenotype suggested that DGAT2 has a very specialized function.

The other important feature that emerged was a key difference between DGAT1 and DGAT2, namely the multi-functionality of DGAT1 (but not DGAT2) in being able to esterify both DAG and MAG [54,55]. This is particularly important not only because it allows DGAT1 to esterify both the glyceride products of TAG hydrolysis, but also because in non-hepatic tissues, MAG may be the major substrate for TAG synthesis (see below). Otherwise, there are only minor differences in the substrate specificities between the two enzymes in vitro, which in vivo are likely to be overwhelmed by the specialization arising from compartmentalization and protein–protein interactions. Thus when Tung Tree epitope-tagged DGAT1 and DGAT2 were expressed in cultured cells, they occupied very discrete non-overlapping regions of the ER [56]. Moreover, plant DGAT2 has been found to interact with GPAT (glycerol-3-phosphate acytransferase) [57], and in animal cells DGAT2 (but not DGAT1) co-localizes with SCD1 (stearoyl-CoA desaturase 1) [58] and a protein with acyl-CoA synthase activity [FATP1 (fatty acid transport protein 1)] [59]. These observations suggest that DGAT2 exists in a multiprotein micro-environment within which substrates and products of fatty acid acylation, desaturation and esterification are channelled sequentially among the closely interacting constituent enzymes.

INTRACELLULAR LOCATION AND MEMBRANE TOPOLOGY OF DGAT1 AND DGAT2 CATALYTIC SITES

The discovery that DGAT activity is expressed on both aspects of the ERM [37] suggested that one of the newly described DGAT proteins would have its active site exposed on the cytosolic aspect of the ERM, and the other on the luminal aspect. However, this turned out to be an oversimplification of the situation. The exclusively overt expression of the DGAT2 active site is now well established through the use of several different approaches. In particular, epitope tagging of heterologously expressed protein from both plant and mammalian sources demonstrated that DGAT2 is a polytopic protein with two TM (transmembrane) domains separated by a short loop, with both the N-terminal and the larger and catalytic site-bearing C-terminal domain exposed on the cytosolic aspect of the ERM [60]. DGAT2 is present not only on the ERM, but also on LDs and mitochondria-associated ER [61]. Its function on LDs appears to be to enable their expansion through its interaction with the ER protein FATP1 [59]. Therefore all DGAT2 activity is ‘overt’, as it can be accessed by cytosolic acyl-CoA esters, although its location is heterogeneous, having been described on the surface of cytosolic LDs and mitochondria. This has been ascertained through inhibitor and transgenic mouse studies in which it was found that activity corresponding to that of DGAT2 is expressed entirely overtly (i.e. is fully accessible to cytosolic acyl-CoA substrates) [62,63].

The topology of DGAT1 is more complex. Whereas epitope-tagging and partial proteolysis approaches concluded that the DGAT1 catalytic site is expressed exclusively on the latent aspect of the ERM [64], several other lines of evidence suggest that the enzyme has a dual topology [62,63], with its active site expressed on both the overt and latent aspects of the membrane. Thus microsomes isolated from Dgat1−/− mice have no latent DGAT activity, which suggests that the latent activity is due entirely to DGAT1 [62,63]. However, in these microsomes there is also a diminution of a substantial fraction of the overt DGAT activity, indicating that DGAT1 activity is also expressed on the overt aspect of the ERM [63]. Moreover, experiments conducted on HepG2 cells with two specific inhibitors developed against human DGAT1, and belonging to different classes of chemicals, showed that overt and latent DGAT1 activities are differentially sensitive to these two classes of pharmacological agents [63]. These observations indicated that DGAT1 has dual membrane topology in the ERM of HepG2 cells, with DGAT1 molecules being orientated in one of two possible conformations within the membrane, one with its active site available to cytosolic substrates, and the other to luminal substrate pools (specifically acyl-CoA). This agreed with the data in [65] where it was observed that there was an increase in both overt and latent DGAT activity when DGAT1 was expressed adenovirally in the liver of mice. Dual topology of DGAT1 in the ERM has also been found in the enterocyte-derived cell line Caco-2 (H.R. Wurie, L. Buckett and V.A. Zammit, unpublished work).

The molecular mechanism through which DGAT1 achieves its dual topology is still being elucidated. However, it may be pertinent that DGAT1 exists as a tetramer [66], because the existence of oligomeric membrane proteins with dual topology is now well established [6770]. Dual topology of oligomeric membrane proteins is associated with weak topographical motifs flanking the boundaries of their TM domains. This allows each individual protomer to adopt an orientation independently of the other protomers within the overall oligomeric structure, resulting in their active sites being expressed on both aspects of the membrane [69]. The ability of at least one enzyme that, in yeast, is able to vary the relative proportion of number of protomers facing one or the other aspect of the membrane depending on the nutrient status of the organism suggests that the balance between the protomers adopting each of the two topologies is regulated by co- or post-translational mechanisms [67,71]. DGAT1 exists as a dimer/tetramer in vivo, and the partitioning of individual protomers between the two possible membrane orientations may be the mechanism through which the dual topology of its active site in the ERM is achieved. Interestingly, changes in the relative magnitudes of overt and latent DGAT activities in response to physiological, dietary and pharmacological (fibrate, statin) treatments have been described [72], suggesting that the relative proportions of the number of protomers in each orientation is under metabolic control. In particular, there is a 5-fold increase in the overt/latent ratio of DGAT activity in rat liver microsomal membranes following fibrate treatment of rats [72]. Considering that DGAT1 only constitutes a fraction of overt DGAT activity in the control state, this increase in overt DGAT activity may be the result of an even greater re-configuration of DGAT1 orientation across the ERM after fibrate treatment. Such redistribution of DGAT1 topology across the ERM to favour cytosolic expression of its active site may contribute towards the beneficial hypolipidaemic effect of fibrates [72] especially as lumen-facing DGAT1 may determine the size of VLDL particles (see below).

Recently, a mechanism has been described for the physiological control/modulation of the relative proportions of protomers of dual-topology proteins that face either aspect of eukaryotic membranes [73,74]. This model proposes that physiologically induced changes in membrane lipid composition may affect the topology of individual conformers both during co-translational insertion and by enabling re-orientation of the protein within the membrane even after its insertion. This would provide the basis for a mechanism through which the topology of such proteins can be affected physiologically by nutritionally and hormonally mediated changes in membrane lipid composition. Such a mechanism may be harnessed pharmacologically as a strategy of altering the fraction of TAG that is secreted by the liver and the TAG content (size distribution) of VLDL.

The ability of DGAT1 to vary its topology in the ERM may also enable it to adapt its orientation to the physiology of the different cell types in which it occurs. Thus although having a topology that results in partial latency of its active site within the ERM is important for the function of cell types that secrete TRLs (hepatocytes and enterocytes), it would be inappropriate for TAG formation in non TRL-secreting cell types (e.g. myocytes and adipocytes) in which TAG-containing LDs are exclusively located in the cytosolic compartment. Therefore, for DGAT1, possessing membrane-insertion motifs that enable it to be expressed with either of the two possible topologies in the same membrane would ensure that DGAT1 activity is expressed appropriately on either aspect of the ERM in different cell types, depending on their acute or long-term physiological requirements.

PRE-REQUISITES FOR HEPATIC TAG SYNTHESIS AND SECRETION

The importance of the dual topology of DGAT1 within the ERM of hepatocytes is that it enables it to maintain distinct TAG pools on both sides of the membrane in the presence of lipase action. The liver contains several lipases that hydrolyse TAG at high rates on both the cytosolic and ER luminal LDs [75]. Therefore the expression of DGAT1 activity on both aspects of the membrane is important to maintain both TAG pools, especially as DGAT1 is the only one of the two DGATs that can utilize both DAG and MAG (and long-chain fatty acyl-CoAs) as substrates [55] (see below). These properties enable DGAT1 to rescue both of these partial glycerides from complete hydrolysis. The question arises however, as to why DGAT2 cannot perform this function on the cytosolic aspect of the ERM, i.e. why DGAT1 activity is required also on the cytosolic aspect of the ERM. The answer appears to lie in the specialized role that DGAT2 plays in the pathway of TAG synthesis.

There are two essential features of TAG synthesis and secretion in the liver that have been recognized for a considerable amount of time, but that appear to be paradoxical. These are (i) their dependence on de novo lipogenesis of fatty acids [7678], and (ii) their dependence on the endogenous desaturation of the products of de novo fatty acid synthesis [79]. The ability of high-carbohydrate diets to activate de novo fatty acid synthesis through the activation of the relevant transcription factors [e.g. SREBP-1c (sterol-regulatory-element-binding protein 1c) and ChREBP (carbohydrate-responsive-element-binding protein)] is well established [80]. This link between de novo fatty acid synthesis and TAG synthesis/secretion was experimentally demonstrated in vivo in mice bearing a hepatocyte-specific deletion of the insulin receptor (LIRKO mice) [81]. These mice had increased apoB secretion, but lower VLDL–TAG secretion. These changes were accompanied by a down-regulation of FAS (fatty acid synthase) and SCD1, but a marked increase in DGAT1 expression (DGAT2 was not measured) [81]. The observation of the down-regulation of SCD1 induced by the absence of insulin signalling is significant, as hepatic SCD1 deficiency results in a markedly impaired ability of the liver to synthesize and secrete TAG [79,82]. Therefore there appears to be a requirement by the liver for fatty acids synthesized and desaturated in situ in order for it to be able to synthesize and secrete TAG. The requirement for desaturation of de-novo-synthesized fatty acids for hepatic lipogenesis and TAG secretion to occur is likely to be due to the effects that the unsaturated products of SCD1 have on the expression of lipogenic enzymes, through their effects on transcription factors (SREBP-1c and ChREBP) involved in carbohydrate-induced activation of de novo fatty acid synthesis [22,79].

However, the dependence of TAG secretion on de novo lipogenesis is not what would be expected from the very minor contribution that de-novo-synthesized fatty acids make to secreted TAG. They constitute a small fraction (<5%) of those secreted within VLDL–TAG (both in animal models and humans) unless the carbohydrate content of the diet is increased substantially [83,84]. Therefore the liver superficially appears to depend on a ‘catalytic’ amount of de-novo-synthesized fatty acid to enable hepatic TAG synthesis and secretion to occur [35,36,85,86]. Whereas the link between carbohydrate intake and hepatic fatty acid synthesis could be rationalized by the insulinaemic effects of carbohydrate-rich diets on lipogenesis, the identity of the link to TAG synthesis and secretion remained unresolved. The answer appears to centre around the specialized and rate-limiting function of DGAT2 [63].

DGAT2 AND DGAT1 ACT IN SERIES AND NOT IN PARALLEL

As mentioned above, DGAT2 activity is entirely ‘overt’, i.e. it is accessible to cytosolic substrates. This agrees with its membrane topology in the ERM [60] and its detection on the cytosolic LDs and mitochondria-associated microsomes [61]. It accounts for approximately 30% of overt DGAT activity [63], and less than 15% of overall cellular DGAT activity [63,87]. And yet, inhibition of such a small fraction of cellular DGAT activity has an almost total inhibitory effect on the de novo synthesis of TAG (defined as the de novo incorporation of the glyceroyl moiety into TAG) [63]. By contrast, inhibition of DGAT2 does not affect the incorporation of exogenously supplied oleate into TAG (which can occur independently of any net synthesis of glyceride due to hydrolysis–re-esterification cycling between TAG and partial glycerides) [63].

Specific inhibition of DGAT1 results in a totally different pattern of loss of incorporation of glycerol and preformed fatty acids into TAG. There is a strictly proportional loss of labelling of both the glyceride and acyl parts of the TAG molecule [63]. Therefore there is an extreme specialization of the two DGATs. DGAT2 is specialized for the incorporation of glycerol 3-phosphate, but not of exogenous preformed fatty acid into TAG. These observations were corroborated by the observations of [87] which showed, using stable isotopes, that DGAT2 inhibition results in the loss of incorporation of glycerol, but not of oleate, into TAG, both in hepatocytes and in vivo. By contrast, DGAT1 appeared to be specialized for incorporation of oleate, but not glycerol, into TAG [87].

These observations raise questions as to (i) the identity of the fatty acid substrates for DGAT2 (since it does not use preformed exogenously supplied fatty acid), and (ii) how the glyceroyl moiety newly incorporated into DAG is brought together with long-chain fatty acids into the same TAG molecule. The possibility emerged that if DGAT2 does not utilize preformed exogenous fatty acids it may be specialized for the utilization of de-novo-synthesized fatty acids that arise endogenously. This was found to be the case [63]. Down-regulation of the relatively low DGAT2 activity [by a selective inhibitor or by siRNA (small interfering RNA)-mediated knockdown] in HepG2 cells had the same disproportionately large (near total) inhibition of the incorporation of acetate-derived fatty acids into TAG. Therefore inhibition of DGAT2 affected glycerol and acetate incorporation into TAG in parallel, indicating that DGAT2 utilizes nascent diglycerides (newly formed and containing de-novo-synthesized fatty acids) and de-novo-synthesized long-chain acyl-CoA as substrates [63]. (‘Nascent diglycerides’ refers to those DAG molecules that arise from de novo incorporation of the glyceroyl moiety, i.e. not those arising from the hydrolysis of TAG.) By contrast, loss of DGAT1 inhibited glycerol and oleate incorporation equally and stoichiometrically (at a 1:3 ratio), suggesting that when DGAT1 activity is inhibited, the loss of any acyl chain from a TAG molecule, due to lipolysis, results in the loss of the entire glyceride entity from the TAG fraction [63].

The parallel effects on (i) glycerol and acetate incorporation by DGAT2 inhibition, and (ii) on glycerol and oleate incorporation by DGAT1 inhibition indicate that the product of DGAT2 generates (after lipolysis) the DAG substrate for DGAT1, i.e. that DGAT2 acts upstream of DGAT1 (Figure 1) [63].

IMPLICATIONS OF THE SEQUENTIAL ACTION OF DGAT2 AND DGAT1

There are very important implications of the observations described above, namely (i) that DGAT1 and DGAT2 act in series rather than in parallel in hepatocytes; (ii) that DGAT2 is primarily responsible for the initial synthesis of TAG, and thus acts upstream of DGAT1; and (iii) that DGAT1 functions primarily to retain acyl groups and the glyceride moiety within the TAG fraction, by catalysing the re-esterification of DAG and MAG formed after lipase-mediated hydrolysis of TAG. The rapid TAG–partial glyceride cycling that occurs in hepatocytes [33,34], a process during which 70% of TAG becomes racemized [35], results in the TAG secreted in VLDL having a different stereo-isomeric composition of acyl chains than of that in cytosolic LDs [35]. The action of DGAT2 upstream of DGAT1, indicates that it has access to pools of de-novo-synthesized diglyceride and fatty acids that are not available to DGAT1. This explains the apparently paradoxical and long-standing observations that VLDL–TAG formation is strictly linked to the rate of de novo fatty acid synthesis in the liver [7678]. Such a rate-limiting role of DGAT2 would not have been discerned from the modest contribution that de-novo-synthesized fatty acids make towards secreted TAG under normal carbohydrate diets [83]. Thus although in both humans and animal models fed a normal carbohydrate diet the absolute contribution of de-novo-synthesized fatty acids to VLDL–TAG is very low (~5%) [83,88], inhibition of fatty acid synthesis results in an almost complete cessation of TAG synthesis and VLDL–TAG secretion [7678]. Similarly, inhibition of SCD1 or disruption of its gene have an almost complete inhibitory effect on hepatic TAG synthesis in vivo [89] which cannot be rescued by dietary triolein feeding [89], suggesting that the DGAT enzyme that utilizes de-novo-synthesized fatty acids exerts a rate-limiting effect on overall TAG synthesis and secretion. This is in agreement with the rate-limiting role predicted from the upstream function of DGAT2 suggested in [63] (Figure 1).

However, a minor, but substantial, proportion of TAG synthesized by DGAT2 does not appear to undergo hydrolysis–re-esterification cycling [63]. This suggests that this portion of newly synthesized TAG is sequestered into a pool(s) that bypasses the hydrolysis–re-esterification cycling and is incorporated directly (i.e. used intact) for incorporation into LDs or for secretion. It is well-established that a sizeable minority of TAG is secreted without prior hydrolysis–re-esterification [33,34]. In this respect, in [35,36] it was shown that in vivo only approximately 70% of newly synthesized TAG is racemized before secretion by the liver. Similarly, studies on LD formation have suggested that different LDs may have differential access to lipogenic enzymes [90], and that DGAT2, but not DGAT1, is specifically associated with LDs [90] and may form part of a bridge to the ERM in association with FATP1 [59]. In this context, it is of interest that others have observed that small ‘new’ and large ‘old’ LDs in the cytosolic compartment of McArdle cells have different dynamics, and that whereas there is transfer of TAG from small nascent droplets to larger LDs, this does not happen in the reverse direction [30].

A role for DGAT1 in ‘rescuing’ partial glycerides (DAG and MAG) after their formation through TAG hydrolysis, means that this enzyme is very important in the retention of glycerides within the TAG fraction of the liver (and other tissues, see below). This may explain why overexpression of DGAT1 in mouse liver in vivo increases hepatic TAG content [91]. DGAT1 has been shown to have considerable MGAT activity in its own right [55], and may, therefore, re-esterify both DAG and MAG to TAG. Therefore DGAT1 is expected to play an essential role in the accumulation of TAG either for retention in the cell or as a source of TAG for secretion. Without its action, DAG and/or MAG generated by lipases could be diverted to phospholipid synthesis (from DAG) or total hydrolysis (from MAG) to glycerol and fatty acids. DGAT1 activity may need to be much higher compared with that of DGAT2 in hepatocytes, as its catalytic activity would need to accommodate the very rapid cycling flux between TAG and partial glycerides [33,34]. Because of the extent of this cycling, the de-novo-synthesized fatty acids incorporated into TAG during initial (DGAT2-mediated) synthesis may be rapidly replaced by preformed fatty acids in the re-esterification reactions catalysed by DGAT1, hence explaining the low content of de-novo-synthesized acyl chains in secreted TAG, unless the rate of de novo fatty acid synthesis is increased substantially, e.g. under conditions of carbohydrate-diet-induced lipogenesis [83].

The model in Figure 1 may also explain the observation, obtained with liver-specific DGAT1-knockout mice, that DGAT1 is required for the development of steatosis due to excess exogenous (preformed) fatty acid supply to the liver, whereas steatosis associated with increased hepatic de novo fatty acid synthesis is not DGAT1-dependent [92]. These data are fully consistent with the conclusion that DGAT2 utilizes substrates that are, or are composed of, de-novo-synthesized (‘new’) fatty acids, whereas DGAT1 is primarily involved in esterifying ‘old’ preformed fatty acids to DAG. More generally, these considerations also explain the extreme phenotype and lethality (soon after birth) of global Dgat2 gene disruption in mice [49], as this would prevent primary TAG synthesis (de novo incorporation of glycerol into TAG). By contrast, global disruption of the Dgat1 gene only leads to a mild phenotype characterized by lower TAG contents of tissues (including the liver) and mild hypotriglyceridaemia [52] as would be expected from a lower rate of ‘rescue’ of DAG and MAG after lipase-mediated hydrolysis of TAG.

The rate-limiting role of DGAT2 for de novo TAG synthesis may explain why niacin is a good hypolipidaemic agent [32]. Niacin inhibits DGAT2 non-competitively, but not DGAT1 [93]. As expected from the relatively low activity of DGAT2 compared with DGAT1 [62], niacin only affects a relatively small percentage of overall DGAT activity [62,63]. It is now evident that by inhibiting DGAT2, niacin would prevent hepatic steatosis associated with endogenous synthesis of fatty acids. Interestingly, the efficacy of niacin as an agent against fatty liver is dependent on DGAT2 polymorphisms of the individuals studied [94], confirming the central role of the enzyme in determining the rate of de novo TAG synthesis and, in turn, of the absolute rates of net de novo TAG synthesis by the liver.

INSULIN EFFECTS ON TAG SYNTHESIS AND SECRETION

Hyperinsulinaemia in vivo is associated with hypertriglyceridaemia [95]. But insulin promotes the degradation of apoB [96], an effect that would act contrary to the lipogenic effect of the hormone [80,97]. In the present context, insulin is expected to increase the rate of DGAT2-mediated TAG synthesis by increasing the rate of de novo lipogenesis through its induction of SREBP-1c [80]. Importantly, this insulin-stimulated lipogenesis is preserved in insulin-resistant states, owing to the phenomenon of selective insulin resistance [81,97]. Moreover, because insulin-resistant states are accompanied by hyperinsulinaemia, hepatic lipogenesis is expected to be stimulated to an even greater extent than normal, rather than being inhibited under these conditions. Thus in mice in which the insulin receptor was specifically knocked out in hepatocytes (LIRKO mice) to produce ‘pure’ insulin resistance, there was a marked decrease in SCD1 and FAS expression and inhibition of VLDL–TAG secretion in spite of an increase in DGAT1 (DGAT2 was not measured) [81]. Moreover, enhanced insulin signalling [mediated by interference with the action of PTEN (phosphatase and tensin homologue deleted on chromosome 10)] increased TAG synthesis and the accumulation of cellular TAG, thus overriding the effect of insulin on apoB degradation, and resulting in increased VLDL–TAG secretion. These observations indicated that the level of both apoB and TAG secretion are determined by the effects of insulin on lipogenic rate and TAG availability rather than by the stability of apoB [98]. This corroborates previous data [88,99] showing that in vivo and in perfused livers (but not in isolated primary hepatocytes) insulin only inhibits TAG secretion by the liver in the fasted state (in rats and humans respectively), and that in the fed insulin-replete state, when lipogenesis is higher, insulin stimulates TAG secretion [100,101]. Therefore the effect of insulin on hepatic TAG secretion is dependent on the prior insulinaemic state of the liver [99]. These observations are reproducible using rat isolated perfused livers [102], but not in isolated hepatocytes unless they are cultured with insulin for a prolonged period [103]. They provide a rationale for the conclusions reached previously, through work on humans fed a high-carbohydrate diet, that hyperinsuinaemic hypertriglyceridaemia is not due to failure of insulin to inhibit TAG secretion [23]. We have previously highlighted the possibility that, due to the stimulation of TAG secretion that it is likely to cause, the possible deleterious effects of premature treatment of newly diagnosed Type 2 diabetic patients with insulin may outweigh any benefits of the strict glycaemic control achieved [100,101], a prospect that has been reiterated more recently [81].

This may now be explained by the link that DGAT2 provides between de novo fatty acid synthesis (which does not become resistant to the action of insulin in the pre-diabetic state [98]) and TAG synthesis. Thus, in the (pre)diabetic state, DGAT2-mediated de novo TAG synthesis proceeds undiminished (if not actually increased), and DGAT1 will have an increased level of substrate available (from circulating fatty acids released from adipose tissue) for re-esterification of partial glycerides. As expected, this combination of increased lipogenesis and increased availability of circulating fatty acids results in the steatosis and hypertriglyceridaemia that accompany this condition, in spite of the relatively increased partitioning of fatty acids towards β-oxidation [104,105]. The effects of the increased glucose and cholesterol output by the liver {owing to insulin resistance of pathways regulated by Akt, but not through mTORC1 (mammalian target of rapamycin complex 1) [81,98,106]}, coupled to the enhanced lipogenesis and associated synthesis of TAG, result in combined hyperglycaemia, hyperinsulinaemia and hypertriglyceridaemia experienced in insulin-resistant states. Therefore DGAT2 emerges as the link between these three important markers of the metabolic syndrome.

THE ROLES OF DGAT2 AND DGAT1 IN NON-HEPATIC TISSUES

Enterocytes

Like hepatocytes, enterocytes secrete TRLs, but the process of TAG synthesis in this cell type is largely one of re-synthesis of TAG from the MAG and fatty acid produced by pancreatic lipase in the lumen of the gut [25]. Therefore there is likely to be only a small amount of net TAG synthesis (from glycerol 3-phosphate and de-novo-synthesized fatty acids). Accordingly, MGAT activity is high in enterocytes. DGAT1 may contribute towards this activity owing to its ability to use MAG as a substrate in addition to its role in catalysing the final reaction in TAG synthesis [55]. In humans, MGAT3 may contribute towards re-esterification in enterocytes owing to its DGAT [107] activity [47,108].

The dual distribution of DGAT1 on the overt and latent aspects of the ER membrane of Caco2 cells has been experimentally demonstrated (H.R. Wurie, L. Buckett and V.A. Zammit, unpublished work) indicating that the role of DGAT1 in re-esterifying partial glycerides within the ER lumen also exists in this cell type. Indeed, although it was at first suggested that DGAT1 is not necessary for dietary TAG uptake [109], it is now known that intestinal TAG absorption in mice lacking DGAT1 specifically in the intestine is delayed, resulting in much lower post-prandial excursions of TAG [110]. This suggests that, although as in the liver the direct transfer of intact TAG during co-translational translocation of apoB across the ERM occurs, in the absence of DGAT1 the high rates of secretion of TAG enabled by resynthesis of TAG within the ER lumen is not possible. Specific expression of DGAT1 solely in enterocytes of Dgat1−/− mice reverses the resistance of the animals to high-fat-diet-induced obesity [111], indicating that the lowering of the post-prandial excursions of TAG may be sufficient to relieve the lipotoxicity of tissues such as skeletal muscle. Treatment of mice with a specific DGAT1 inhibitor totally abolishes post-prandial TAG excursions and the appearance of labelled acyl groups from dietary triolein in chylomicrons [110]. Interestingly, although they had similar effects on chylomicron TAG secretion, respective inhibition of MTP and of DGAT1 had opposite effects on TAG accumulation in enterocytes [110]. Thus MTP inhibition resulted in the accumulation of cellular TAG, whereas DGAT1 inhibition resulted in a decrease in cellular TAG content [110]. This is likely to have arisen from the fact that inhibition of MTP results in the backing-up of TAG within the cell when the MTP-dependent mechanisms for TAG incorporation into chylomicrons are lost, especially since TAG continues to be absorbed normally [110]. In contrast, when enterocyte DGAT1 is inhibited, the (re)synthesis of TAG from partial glycerides is not possible, resulting in total hydrolysis of the partial glycerides to fatty acids and glycerol.

In enterocytes, the predominant use of sn-2-monoglyceride for intracellular TAG (re)synthesis suggests that DGAT2 would be mostly bypassed, as only DGAT1 (and other enzymes with MGAT activity) would be required; DGAT2 expression, like de novo fatty acid synthesis, is low in enterocytes [109] (see Figure 2). However, TAG secretion by the intestine is positively related to the rate of de novo fatty acid synthesis, although it is not dependent on it [107]. This indicates that while the rate of de novo fatty acid synthesis in enterocytes is likely to be significantly lower than that in the liver, DGAT2 may still be able to play a role in linking the two processes under conditions in which lipogenesis from glucose (e.g. in diabetes) is elevated; it may be involved in promoting the increased rate of chylomicron TAG secretion observed in the fructose-fed hamster, an animal model of diet-induced insulin resistance [107,112].

Figure 2 Comparison of the possible routes to TAG synthesis in hepatocytes, enterocytes and myocytes

In tissues in which, unlike the liver, de novo synthesis of fatty acids from carbohydrate sources is not a major process (intestine and muscle), DGAT2 may be bypassed through the formation of partial glycerides by lipase-mediated hydrolysis of TAG (e.g. by pancreatic lipase and LPL). These glycerides can then be used by DGAT1 to resynthesize TAG within the respective cell types (enterocytes and myocytes). When de novo fatty acid synthesis from glucose becomes more important (e.g. in hyperglycaemiac and hyperinsulinaemic states in muscle) it would provide substrates for DGAT2. In adipocytes, the involvement of DGAT2 may be more significant, as lipogenesis from glucose may be more prevalent. FA, fatty acid.

Myocytes and adipocytes

Cardiac and skeletal muscle fibres can accumulate substantial amounts of TAG [24,113]. In the heart, this is associated with cardiomyopathy [113], and in skeletal muscle with insulin resistance when it occurs under conditions of dietary excess [79,114]. However, muscles of athletes (which can oxidize fatty acids at high rates) also have a high intra-myocytic TAG content [115,116], a phenomenon termed the ‘athletes paradox’. Myocytes receive a substantial part of their fatty acids as the products of LPL (lipoprotein lipase) action on VLDL– and chylomicron–TAG. Because LPL activity may result in the incomplete lipolysis to MAG [117] it is possible that, as in enterocytes, when metabolizing the products of LPL action, muscle DGAT2 is bypassed in the formation (resynthesis) of TAG inside the cells (see Figure 2). No MAG is detected in the circulation when chylomicrons are hydrolysed by a subcutaneous depot of adipose tissue in vivo [118], suggesting that MAG formed by LPL may be very efficiently taken up by tissues. Once inside the cells, it is either further hydrolysed by monoglyceride lipase or used directly to synthesize DAG and TAG, in a manner analogous to the use of 2-monoglyceride in enterocytes [25]. In this instance, the MGAT-like activity of DGAT1 may be important in the ‘clearance’ of MAG, DAG and fatty acid (acyl-CoA) from the cytosol of myocytes under dyslipidaemic conditions. DGAT1 would undertake the resynthesis of TAG intracellularly from these products of LPL action. In the process, this would ensure that the concentrations of acyl-CoA and other intermediates (DAG and ceramide) that may induce insulin resistance [15,119] are kept low. This may explain why overexpression of DGAT1 in type-1 muscle fibres renders them more insulin sensitive [120], and would be in agreement with the observation in some studies that exercise acutely increases DGAT1 expression in muscle [8]. However, the correlation of DAG with insulin resistance in muscle is still uncertain. Thus whereas chronic [121,122] or acute [8] exercise lowers cellular ceramide, athletes tend to have a higher intra-myocytic DAG content than sedentary insulin-resistant individuals [122].

Interestingly, overexpression of DGAT2 in type-2 muscle fibres has the opposite effect than that of the expression of DGAT1 in type-1 fibres; it makes muscle more insulin resistant [113], even though the accumulation of TAG in these fibres occurs to the same extent as in type-1 fibres overexpressing DGAT1. Although the targeting of a different fibre type in these experiments complicates the interpretation of these data, they raise the interesting prospect that DGAT2 activity is linked to lipogenesis and the net new synthesis of glycerides also in muscle cells. TAG formed by DGAT2 would be metabolically distinct from that formed by DGAT1. Rather than clearing acyl-CoA, de novo synthesis of new TAG by DGAT2 would provide substrate for lipases that generate the intermediates (acyl-CoA and DAG) that promote insulin resistance (see Figure 3). It is well-established that the intramyocellular content of TAG (IMTG) is high in obese/diabetic individuals, in whom muscle is insulin resistant, but also in trained athletes, in whom it is very insulin sensitive. In view of the specialized functions of DGAT2 and DGAT1, it is possible that this ‘athletes paradox’ [115,116] may be related to the respective properties of the distinct pools of TAG synthesized by DGAT1 (with its ability to remove lipid intermediates) or DGAT2 (linked to lipogenesis-driven synthesis of new TAG). Although muscle is not normally thought to be a lipogenic tissue, even under basal conditions it uses a considerable fraction of glucose uptake (~5%) to synthesize fatty acids, and this proportion increases markedly when myocytes are cultured under hyperglycaemic conditions in vitro [123,124]. Moreover, in myotubes cultured from human muscle satellite cells, hyperglycaemia induces SREBP-1c expression and lipogenesis, with a concomitant increase in TAG content [123,124]. Therefore the specialization and sequential nature of DGAT2 and DGAT1 activities may explain why both insulin-resistant and insulin-sensitive muscle can have equivalent amounts of overall cellular TAG, but have opposing phenotypes (and different LD morphology). It is of note that overexpression of DGAT1 in hearts of mice transgenic for PPARγ (peroxisome-proliferator-activated receptor γ) rescues the myopathic phenotype of the animals without decreasing the overall TAG content of the cardiomyocytes, the only observable difference between the two phenotypes being the change in morphology of the LDs in cardiac myocytes from small to large [5].

Figure 3 Proposed roles of the specialized functions of DGAT1 and DGAT2 in the alleviation and promotion respectively, of insulin resistance in skeletal and cardiac muscle

Both insulin-resistant (e.g. in Type 2 diabetic) and highly insulin-sensitive (e.g. in athletes) muscle cells contain high levels of TAG. Besides differences in the rate at which these muscles can oxidize fatty acids, the DGAT enzyme responsible for the synthesis of TAG may be involved. (A) In exercised muscle, increased DGAT1 activity clears acyl-CoA and DAG to form TAG from preformed fatty acids. (B) In obese/diabetic individuals, increased lipogenesis induced by hyperglycaemia and hyperinsulinaemia provides substrates for DGAT2. The TAG thus synthesized is a source of acyl-CoA and DAG generated by intracellular lipases. ATGL, adipose TAG lipase; FA, fatty acid; HSL, hormone-sensitive lipase; IS, insulin sensitivity; NEFA, non-esterified fatty acid.

In adipocytes, de novo fatty acid synthesis from glucose may be an important process depending on the species and the fat content of the diet. Therefore DGAT2 would be expected to be important in the function of the tissue. Accordingly, both DGAT1 and DGAT2 were found to be important for TAG synthesis and LD formation in adipocytes [50]. In differentiating adipocytes, the induction of DGAT2 was found to be dependent on the activation of C/EBP (CCAAT/enhancer-binding protein) β [125] and to be maintained by C/EBPα, after initial differentiation, indicating a central role for the enzyme in adipose tissue function.

HEPATIC DGAT2 AS THE LINK BETWEEN GLYCAEMIA AND TRIGLYCERIDAEMIA

It is well established that a crucial factor in driving hypertriglyceridaemia is insulin-stimulated hepatic de novo lipogenesis mediated by increased nuclear SREBP-1c [80,97]. In this respect, DGAT2 emerges as the link between glycaemia (and its attendant insulinaemia) and the net new formation of TAG. It occupies a fundamentally important node at which carbohydrate, fatty acid and TAG metabolism intersect (see Figure 4). Its relatively low activity within hepatocytes is rate-limiting for the de novo synthesis of TAG. The interaction of DGAT2 with an expanding list of other proteins [SCD1, GPAT and ACS (acyl-CoA synthetase)–FATP1, see above] is probably required to create its own microcompartment in the cell, within which newly synthesized fatty acids are channelled towards the de novo synthesis of DAG and of DGAT2-catalysed TAG synthesis. Therefore, through DGAT2 action, the lipogenesis that is stimulated by insulin (via SREBP-1c) and carbohydrate (via ChREBP) is linked to de novo TAG synthesis. Together with the function of DGAT1 of salvaging partial glycerides from complete hydrolysis, DGAT2 action in the liver is expected to result in hepatic steatosis, hypersecretion of VLDL–TAG and the provision of substrate for the formation of molecular species that are associated with insulin resistance.

Figure 4 The role of DGAT2 as the link between glycaemia and triglyceridaemia

Hyperglycaemia and attendant hyperinsulinaemia stimulate de novo fatty acid synthesis through the activation of transcription factors including SREBP-1c and ChREBP. DGAT2 utilizes the products of de novo fatty acid synthesis and of de novo DAG formation for the synthesis of TAG which is used (after hydrolysis and re-esterification by DGAT1) for TAG secretion within VLDLs. Therefore DGAT2 links hyperglycaemia (and attendant hyperinsulinaemia) and hypertriglyceridaemia.

Therefore DGAT2 plays a crucial role in integrating carbohydrate and lipid metabolism in the liver owing to the ability of the tissue to synthesize both fatty acids and TAG de novo, and to use the products of both of these processes as substrates for the enzyme. Inhibition of DGAT2 pharmacologically in the liver would be expected to break this link and to prevent the carbohydrate-mediated steatosis, insulin resistance and hypertriglyceridaemia observed under hyperinsulinaemic conditions. The selective insulin resistance [80,81] which enables both increased glucose output from the liver concomitantly with increased lipogenesis places DGAT2 at the centre of the glucose and fatty acid disturbances of the metabolic syndrome and Type 2 diabetes.

FUNDING

The work of the author is supported by the Medical Research Council (UK), Diabetes UK, the British Heart Foundation and the Biotechnology and Biological Sciences Research Council (UK).

Acknowledgments

I thank the many research colleagues who have contributed to the work described in the present review.

Abbreviations: apoB, apolipoprotein B; C/EBP, CCAAT/enhancer-binding protein; ChREBP, carbohydrate-responsive-element-binding protein; DAG, diacylglycerol; DGAT, DAG acyltransferase; ER, endoplasmic reticulum; ERM, ER membrane; FAS, fatty acid synthase; FATP1, fatty acid transport protein 1; GPAT, glycerol-3-phosphate acytransferase; IMTG, intramyocellular triacylglycerol; LD, lipid droplet; LDL, low-density lipoprotein; LPL, lipoprotein lipase; MAG, monoacylglycerol; MGAT, MAG acyltransferase; MTP, microsomal transfer protein; SCD1, stearoyl-CoA desaturase 1; SREBP-1c, sterol-regulatory-element-binding protein 1c; TAG, triacylglycerol; TM, transmembrane; TRL, TAG-rich lipoprotein; VLDL, very-LDL

References

View Abstract