|
|
|
|
|
|
|
|
|||||
| Home Help Feedback Subscriptions Archive Search Table of Contents | |||||
First published online May 24, 2006
doi: 10.1242/10.1242/jcs.02941
Research Article |
1 Wallenberg Laboratory for Cardiovascular Research, Göteborg University, Sahlgrenska University Hospital, SE-413 45 Göteborg, Sweden
2 Department of Pharmacological Science and the Center for Developmental Genetics, Stony Brook University, Stony Brook, New York 11794, USA
Author for correspondence (e-mail: Sven-Olof.Olofsson{at}wlab.gu.se)
Accepted 15 February 2006
 |
Summary |
|---|
 Top Summary IntroductionResultsDiscussionMaterials and MethodsReferences |
|---|
or p38ß were without effect. Insulin stimulated the formation of lipid droplets and this stimulation was inhibited by knockdown of PLD1 (by siRNA) and by inhibition or knockdown (by siRNA) of ERK2. Inhibition of ERK2 eliminated the effect of PLD1 on lipid droplet formation without affecting PLD1 activity, suggesting that PLD1 functions upstream of ERK2. ERK2 increased the phosphorylation of dynein which increased the amount of the protein on ADRP-containing lipid droplets. Microinjection of antibodies to dynein strongly inhibited the formation of lipid droplets, demonstrating that dynein has a central role in this formation. Thus dynein is a possible target for ERK2.
Key words: Phospholipase D1, Extracellular signal-regulated kinase 2 (ERK2), Cytosolic lipid droplets, Insulin, Dynein, Phosphorylation, Adipocyte differentiation-related protein (ADRP)
|
Introduction |
|---|
|
TopSummaryIntroductionResultsDiscussionMaterials and MethodsReferences |
|---|
; Lu et al., 2001). Perilipins are the major PAT proteins in adipocytes and steroidogenic cells (Londos et al., 1999), whereas ADRP is expressed ubiquitously with high levels in the liver (Brasaemle et al., 1997). Several other proteins have been identified on droplets. Among those are caveolin (Fujimoto et al., 2001; Ostermeyer et al., 2001; Pol et al., 2001), which associates with newly formed droplets (Marchesan et al., 2003).
Excessive accumulation of triglycerides in ADRP-containing lipid droplets, particularly in the liver and skeletal muscle, is associated with metabolic disorders such as insulin resistance and type 2 diabetes (Browning and Horton, 2004; den Boer et al., 2004), which are strong risk factors for cardiovascular disease. Moreover, ADRP is the major PAT protein found on lipid droplets in the lipid-loaded macrophages (`foam cells') that are landmarks of atherosclerotic lesions (Chen et al., 2001).
Lipid droplets consist of a core of neutral lipids surrounded by a monolayer of amphipathic lipids, such as phospholipids, to which proteins are bound (Brown, 2001; Murphy and Vance, 1999). The lipid droplets form from microsomes (Marchesan et al., 2003), commencing with the `oiling out' of the triglycerides between the leaflets of the microsomal membranes, generating a lens that buds from the membrane to create cytosolic lipid droplets (Brown, 2001; Murphy and Vance, 1999). Consistent with this model, cytosolic lipid droplets contain the microsomal membrane protein caveolin (Fujimoto et al., 2001; Marchesan et al., 2003; Pol et al., 2001) and proteins from the microsomal lumen (e.g. GRP 78) (Marchesan et al., 2003; Prattes et al., 2000). The primordial particles formed from the microsomes have a diameter of around 0.1 µm and contain ADRP, vimentin, caveolin and GRP78 (Marchesan et al., 2003). These particles increase in size by a fusion process that is dependent on microtubules and motor proteins. One of these, dynein, has been identified on the ADRP-containing droplets (Bostrom et al., 2005).
Using a cell-free system, we have previously described two key factors in the formation of lipid droplets: phospholipase D (PLD), and an unknown cytosolic protein (Marchesan et al., 2003). In the present report, we describe identification of PLD1 as the isoform involved in this process, and we present data implicating ERK2 (extracellular signal-regulated kinase 2) as the unknown cytosolic protein. Moreover, we demonstrate that PLD1 and ERK2 are essential in the increase of lipid droplet formation that is triggered by insulin stimulation. The activity of ERK2 is shown to be essential for the PLD1-mediated increase in lipid droplet formation, but it does not affect PLD1 activity, suggesting that it functions as a PLD1 effector or in parallel with PLD1. Finally, we demonstrate that ERK2 increases the phosphorylation of dynein, which is important in lipid droplet assembly. Thus, the phosphorylation of dynein increases its association with ADRP-containing lipid droplets, and dynein is essential for assembly of the droplets.
|
|
Results |
|---|
|
TopSummaryIntroductionResultsDiscussionMaterials and MethodsReferences |
|---|
). To determine which isoform is responsible for this effect, we examined the consequences of overexpression of PLD1 (Hammond et al., 1995) and PLD2 (Colley et al., 1997) on lipid droplet formation in NIH 3T3 cells. Using immunohistochemistry, we confirmed that PLD1 is located in the interior of the cell (in the cytosol, ER and Golgi structures) whereas PLD2 is confined to the plasma membrane (supplementary material Fig. S1), which is consistent with previous reports (Colley et al., 1997; Du et al., 2003; Du et al., 2004; Freyberg et al., 2001).
Transfection of NIH 3T3 cells with a PLD1-expressing plasmid resulted in a 16-fold increase in the amount of PLD1 in the cell (as determined by western blot and quantification using an anti-PLD1 antibody; data not shown). Elevation of the level of PLD1 expression was found to increase the total area of Oil Red O-stained cytosolic lipid droplets per cell (Fig. 1A,B), in comparison with cells transfected with plasmid expressing a catalytically inactive allele of PLD1 (K898R) (Sung et al., 1997) (the expression of the mutant gave rise to the same increase in PLD1 protein as the expression of the wild-type PLD) or empty control vector. Transfection of PLD1 wild type (wt) led to a 5.1±2.6-fold increase (mean ± s.d.; n=7; P<0.001, Mann-Whitney rank sum test) in the total area of Oil Red O-stained lipid droplets per cell as compared to the inactive mutant. In each individual experiment there was a highly significant difference between the PLD1wt and the mutant [P=0.008 (one experiment) and P<0.001 (six experiments)]. When compared to an empty vector, PLD1wt gave rise to a 2.5±1.8-fold increase (mean ± s.d.; n=5; P=0.008, Mann-Whitney rank sum test). Each of these five experiments showed highly significant increases: P<0.001 (three experiments), P=0.002 (one experiment) and P=0.007 (one experiment) (Fig. 1B). When compared to an empty vector, transfection with PLD1(K898R) resulted in a decrease in the formation of lipid droplets. The difference was not significant (P=0.084) but could indicate a limited dominant-negative effect of the PLD1 K898R on the assembly of lipid droplets. Consistent with the increase in total area of Oil Red O-stained lipid droplets, PLD1 overexpression increased the accumulation of triglycerides (Fig. 1C). PLD1 overexpression also increased the level of expression of ADRP (Fig. 1D; four different experiments with the same result have been carried out).
Transfection of a PLD2-expressing plasmid resulted in a 20-fold increase in PLD2 protein (as determined by western blotting and quantification using an anti-PLD2 antibody, data not shown). However, PLD2 overexpression did not increase the total area of Oil Red O-stained lipid droplets (Fig. 1E) when compared to transfection of the empty vector (fold increase 0.8±0.1; n=3) or an inactive form of PLD2 (PLD2-K758R; fold increase 1.1±0.06; n=3), nor did PLD2 overexpression affect the level of expression of ADRP (Fig. 1F, three different experiments showed the same result). These findings indicate that PLD1, but not PLD2, regulates the intracellular formation of lipid droplets.
To further confirm the effect of PLD1 on the formation of lipid droplets, NIH 3T3 cells (treated with oleic acid to increase the basal production of lipid droplets) were transfected with siRNA to PLD1, which greatly reduced the expression of PLD1 (Fig. 2A). No reduction in the level of expression was observed for the control proteins GRP78 and ß-actin (Fig. 2A). Six different experiments based on SiRNA-mediated depletion of PLD1 showed highly significant reductions (58±16%; mean ± s.d.; P=0.002, Mann-Whitney rank sum test) in the total area of Oil Red O-stained cytosolic lipid droplets per cell. Each of these six experiments showed a highly significant decrease in the total area of Oil Red O-stained lipid droplets per cell when siRNA to PLD1 was compared to control siRNA (P<0.001, in four experiments; P=0.020, in one experiment; P=0.015, in one experiment; Fig. 2B,C). Also, the accumulation of triglycerides (Fig. 2D) and ADRP (Fig. 2E) decreased in a similar way (four different experiments, all showing the same results). Taken together, these results indicate that PLD1 has an important role in lipid droplet assembly in intact cells and that its level of activity is rate-limiting in this process.
|
; Zhang et al., 1999) activated by many types of agonists including insulin (Huang et al., 2005), we investigated whether insulin stimulation would increase the formation of lipid droplets and, if so, whether the response is PLD1 dependent. Insulin provoked lipid droplet formation in serum-starved cells (0.1% calf serum for 12 hours) in a dose-dependent manner, with maximal responsiveness observed at 3 nM after a 1-hour incubation (165±132% increase in Oil Red O-stained lipid droplets per cell (n=4; P=0.047, t-test). Depletion of PLD1 through siRNA inhibited the insulin-induced lipid droplet formation (Fig. 3A). Thus the insulin-induced lipid droplet formation was reduced by 138±43% (mean ± s.d.; n=3; P=0.005). The effect of insulin treatment was defined as the insulin-induced increase (above buffer control) in the amount of Oil Red O-stained lipid droplets in cells transfected with the control siRNA. These results indicate that siRNA to PLD1 at least eliminates the entire increase induced by insulin. Thus, PLD1 has an important role in mediating insulin-signaled storage of lipid in such droplets.
|
Identification of an activator of lipid droplet formation previously identified using a cell-free system
We have previously reported that one additional signaling component in the PLD pathway regulating lipid droplet formation was an unknown cytosolic factor (Marchesan et al., 2003). A chromatographic procedure to partially purify the cytosolic activator was developed. The procedure is based on a combination of gel- and ion-exchange chromatography as described previously (Marchesan et al., 2003) and hydrophobic interaction chromatography (Fig. 4A). The factor was followed during the enrichment procedure by its ability to activate the formation of lipid droplets from microsomes (Marchesan et al., 2003). Most of the activity was recovered between 62 and 0 mM (NH4)2SO4 (fraction 3 in Fig. 4A,B) using hydrophobic interaction chromatography. This fraction was subjected to SDS-PAGE. The gel was stained with Sypro Ruby (Fig. 4C) and the bands cut out, digested with trypsin and analyzed using matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) and mass spectrometry (MS/MS). A number of proteins were identified (Fig. 4C): band 1, xanthine dehydrogenase/oxidase; band 2, serum albumin; band 3, a mixture of serum albumin and catalase; band 4, ATP citrate lyase; band 5, ERK2; and band 6, adiponectin.
|
Xanthine dehydrogenase, serum albumin and catalase were found abundantly in fractions that lacked activator activity (not shown); thus, these proteins were not investigated further. ATP citrate lyase, adiponectin and ERK2 were assessed for their ability to activate lipid droplet formation in the cell-free system, using recombinant adiponectin and ERK2, and purified bacterial citrate lyase. Citrate lyase and adiponectin did not promote lipid droplet formation, whereas ERK2 showed a dramatic effect [12±2-fold higher activity than the S-fraction (mean ± s.d.; n=3; Fig. 4D)]. This result suggested that ERK2 might be the unidentified cytosolic factor working in conjunction with PLD1 to promote lipid droplet formation.
ERK2 regulates the formation of lipid droplets in intact cells
Transfection of NIH 3T3 cells with a wild-type ERK2-expressing plasmid increased the formation of cytosolic lipid droplets 3.0±1.9 fold (mean ± s.d; n=5; P=0.008, Mann-Whitney rank sum test) when compared to a vector encoding GFP as control. Each of these five different experiments showed a highly significant difference between ERK2 and the control vector encoding GFP (P<0.001, three experiments; P=0.002, one experiment; P<0.005, one experiment) in the area of Oil Red O-stained droplets/cell (Fig. 5A,B). A significant increase in the total amount of triglyceride (Fig. 5C) and the level of expression of ADRP (39±19 per cent; n=3; Fig. 5D) was also observed.
|
|
|
), which inhibits phosphorylation of ERK2 and thereby its activation, resulted in a significant decrease in the formation of Oil Red O-stained lipid droplets, both in the presence (supplementary material Fig. S2) and absence of oleic acid (not shown). The mean level of reduction was 59±14% (n=5).
Lack of effect after manipulation of other MAP kinase family members
The ERK2 siRNA did not affect the level of expression of ERK1 (supplementary material Fig. S3), indicating that endogenous levels of ERK1 are not sufficient to promote lipid droplet formation in the absence of ERK2. To examine whether ERK1 might nevertheless function in this process, we transfected the cells with ERK1-YFP, which resulted in a severalfold increase in the level of expression of ERK1 (supplementary material Fig. S4A). However, the overexpression of ERK1 failed to alter the amount of lipid droplet formation in the cell (supplementary material Fig. S4B). We also employed two different siRNAs to ERK1, to examine the requirement for the kinase. Both significantly reduced the ERK1 expression level (supplementary material Fig. S4C shows a representative siRNA experiment) without any effect on expression levels of control proteins (supplementary material Fig. S4C). However, decreasing the ERK1 expression level did not cause a decrease in the amount of lipid droplets in the cell (supplementary material Fig. S4D).
We also investigated the effect of JNK1 and JNK2 and p38 and p38ß on lipid droplet formation using siRNA and inhibitors. SiRNA to p38 or p38ß (supplementary material Fig. S5A) failed to affect the formation of lipid droplets (supplementary material Fig. S5B), as was the case for an inhibitor to p38 (not shown). SiRNAs to JNK1 or JNK2 (supplementary material Fig. S5C,D) also had no effect, as was the case for JNK inhibitor (supplementary material Fig. S5E).
Taken together, these results suggest that the role of ERK2 in promoting lipid droplet formation is specific to this MAP kinase family member.
Effect of ERK2 on insulin-stimulated lipid droplet formation
ERK2 is well known as a downstream effector of insulin signaling (Carel et al., 1996). To address the potential role of ERK2 in insulin-stimulated lipid droplet formation, we employed the MAP kinase inhibitor Ste-Mek113 (see above and supplementary material Fig. S2), and found that the insulin-induced increase in the formation of lipid droplets decreased (Fig. 3B). Thus insulin-induced increase in the lipid droplet formation decreased by 73±40% (n=5, P=0.004) when the cells were treated with Ste-Mek113. SiRNA to ERK2 gave a similar result; a 114% decrease was observed, again suggesting that inhibition of ERK2 inhibits the insulin-stimulated rate of lipid droplet formation. Together, these results indicate that ERK2 is a key player in insulin-stimulated lipid droplet formation.
The relation between ERK2 and PLD1
The ERK2 inhibitor Ste-Mek113 also completely inhibited the PLD1-induced increase in lipid droplet formation (Fig. 8A). Thus in three different experiments Ste-Mek113 caused a 92±17% (mean ± s.d.; P<0.001) decrease of the PLD1-induced increase in the amount of lipid droplets, indicating that ERK2 is an essential component of the PLD1 signaling pathway. However, Ste-Mek113 did not significantly alter the level of PLD1 activity (97±14% of control activity; n=5). Moreover, ERK2 overexpression did not stimulate PLD1 activity (data not shown). These findings indicate that ERK2 functions downstream of, or in parallel with, PLD1 activation.
|
Results obtained in the cell-free system and in intact cells suggested that ERK2 influences the formation of the lipid droplets from microsomes, but does not affect the rate of triglyceride biosynthesis (supplementary material Fig. S6). We therefore investigated whether ERK2 could be involved in the phosphorylation of proteins present on lipid droplets or associated with their formation. We examined ADRP, caveolin, vimentin, PLD1 and dynein. Homogenates of NIH 3T3 cells were incubated with active ERK2 and fractionated with the PhosphoProtein purification kit. This column binds all forms of phosphorylated proteins (i.e. both serine-, tyrosine- and threonine-phosphorylated proteins). The retained fraction containing the phosphorylated proteins was blotted against antibodies to the proteins of interest. The results showed that ERK2 phosphorylated dynein (detected with antibodies to the intermediate chain) but led to a decrease in phosphorylation of caveolin (Fig. 8B, n=2). There was no effect on any other protein examined. Since dynein appeared to be a potential substrate for ERK2 we focused on this protein.
Phosphorylation of dynein is a key element in the regulation of its function as a motor protein and how it interacts with the correct cargo (King, 2000); dynein is thought to have a potentially central role in the assembly of lipid droplets. For example, we previously demonstrated that intact microtubules are essential for the fusion process by which lipid droplets grow in size (Bostrom et al., 2005). Moreover, we demonstrated that dynein is present on ADRP-containing lipid droplets and that vanadate, which inhibits dynein, also reduces lipid droplet formation (Bostrom et al., 2005).
We first investigated whether phosphorylated dynein is present on lipid droplets. To do this, NIH 3T3 cells were incubated with oleic acid and the ADRP-containing lipid droplets were isolated by gradient ultracentrifugation after nitrogen cavitation (see supplementary material Fig. S7) and fractionated using a PhosphoProtein purification kit. The retained (phosphorylated) and unretained (non-phosphorylated) proteins were analyzed by anti-dynein immunoblotting (Fig. 8C). Virtually all of the dynein present on lipid droplets was retained by the column, indicating that it is phosphorylated.
To investigate whether ERK2 might influence the association between dynein and ADRP-containing lipid droplets, NIH 3T3 cells were transfected with ADRP fused to a His affinity tag (Bostrom et al., 2005), incubated with oleic acid and lysed. The cell lysate was either incubated with active ERK2 (`ERK2') or buffer (`Control') and the ADRP-HAT-containing structures were precipitated with Dynabeads, as described previously (Bostrom et al., 2005). The results demonstrated a large increase in the amount of dynein in the precipitate when cells were incubated with active ERK2 (Fig. 8D). Together, the results indicate that ERK2 can increase the interaction between dynein and ADRP-containing lipid droplets.
To address experimentally the possibility that dynein may influence the assembly of lipid droplets, 3T3 NIH cells were microinjected with either a monoclonal antibody directed against the dynein intermediate chain [this antibody has been used previously to inhibit dynein (Burakov et al., 2003)] or a control immunoglobulin. The results (Fig. 8E) showed that the antibody to dynein caused a significant decrease in the amount of lipid droplets in the cell (P<0.001, P=0.048 and P<0.001 in three different experiments, Mann-Whitney rank sum test). The percentage decrease in these three experiments was 63±16% (mean ± s.d.; P=0.002, t-test).
We next investigated whether dynein is important for the fusion of lipid droplets. To accomplish this, cells were microinjected with the antibody to the intermediate chain of dynein or the control immunoglobulin. The droplets were then stained with Bodipy and followed using confocal microscopy in time-lapse studies as described previously (Bostrom et al., 2005). BioPix software was used to tabulate all fusion events during a 5-minute period [as defined in Bostrom et al. (Bostrom et al., 2005)]. The results (Fig. 8F) indicated that the anti-dynein antibody significantly reduced the frequency of fusions (P<0.001; n=16 for anti-dynein, n=13 for control immunoglobulin). Together, these results indicate that dynein is an important player in lipid droplet formation, and a potential mediator of the ERK2 effect on the assembly of lipid droplets.
The effect of ERK2 on the biosynthesis of triglycerides in the cell
The partially purified activator increased the accumulation of triglycerides in the d
1.055 g/ml fractions in the cell-free system (Marchesan et al., 2003), but it did not affect the rate of biosynthesis of triglycerides in this system. Thus, while there was an increase from 2.3±2.7% (mean ± s.d.; n=10) to 19.6±6.6% (n=5) of the radioactive triglycerides recovered in the d1.055 g/ml density range (i.e. the lipid droplets), the biosynthesis of triglycerides was not affected. Thus, the ratio between the production of triglycerides in the cell-free system incubated with and without the partially purified activator was 1.0±0.62 (mean ± s.d.; n=5). In agreement with this, we found no effect of ERK2 siRNA on the incorporation of [3H]palmitate into triglycerides in NIH 3T3 cells (supplementary material Fig. S6A). Overexpression of ERK2 did not give rise to any significant change in the rate of ß-oxidation (supplementary material Fig. S7A). Together, the results obtained in the cell-free system and in the intact cells indicate that ERK2 influences the formation of lipid droplets from the microsomes but not the rate of triglyceride biosynthesis.
Neither overexpression nor knockdown of PLD1 influences the phosphorylation of ERK2
To address the question of whether PLD1 might influence the activity of ERK2, we investigated the effect of PLD1 on the phosphorylation of ERK2. Cells were transfected with PLD1 or a control vector encoding GFP (`Control') and the effect on the amount of phosphorylated ERK2 and GRP 78 (as control protein) was investigated (supplementary material Fig. S8A). The results showed no obvious difference in the amount of phosphorylated ERK2 (or control protein). We also investigated the effect of PLD1 siRNA on the amount of phosphorylated ERK2 (supplementary material Fig. S8B). Compared with a control siRNA, we could not detect any obvious effect of the knockdown of PLD1 on the amount of phospho-ERK2 (or on the amount of control protein). These results indicate that the amount of PLD1 does not influence the phosphorylation (and activity) of ERK2.
|
Discussion |
|---|
|
TopSummaryIntroductionResultsDiscussionMaterials and MethodsReferences |
|---|
). In this paper, we have demonstrated that PLD1 is the relevant isoform, we have identified the cytosolic protein as ERK2, and we have shown that it regulates the assembly of lipid droplets in intact cells. We have also provided information indicating that the function appears to be specific to ERK2, since no effect on lipid droplet assembly was observed after manipulating ERK1, JNK1 or JNK2, or p38 or p38ß, in terms of presence or activity. We have demonstrated that PLD1 and ERK2 are essential, both for basal generation of lipid droplets and for the augmented production that is stimulated by insulin. Finally, we have presented results suggesting that ERK2 influences the formation of lipid droplets - at least in part - by phosphorylating dynein, a protein that was shown to be essential for the formation of lipid droplets.
Our results indicate that PLD1 (but not PLD2) is involved in regulation of the degree of cytosolic lipid droplet formation. This difference may reflect the localization of the two enzymes, or the way in which they are regulated. PLD1 frequently associates with the intracellular membranes (the Golgi and ER) (Colley et al., 1997; Du et al., 2003; Freyberg et al., 2001), which appear to be the site of biosynthesis of the lipid droplets (Marchesan et al., 2003), whereas PLD2 is located on the inside of the plasma membrane (Colley et al., 1997; Du et al., 2004). Similar patterns of localization were found in this study.
The role of PLD1 in the assembly of lipid droplets is presumably to provide phosphatidic acid (PA) to the process, since results from the cell-free system demonstrated that PA strongly promotes the formation of such droplets (Marchesan et al., 2003). The effect of PA may be related to its ability to influence the structure of membranes (Huang et al., 2005) and facilitate the budding reaction (Kooijman et al., 2003), or it may recruit or activate other proteins involved in the process (e.g. Honda et al., 1999). Phosphatidic acid has been proposed to recruit proteins of importance in the formation of transport vesicles (Manifava et al., 2001). Finally, PA can also be converted to other bioactive lipids such as diacylglycerol and lysophosphatidic acid (Jenkins and Frohman, 2005); any or all of these mechanisms may act in conjunction to mediate the effects observed.
Interestingly, PLD1 is well known as a signaling enzyme that transduces G-protein-coupled and tyrosine kinase receptor-mediated stimuli (Carel et al., 1996; McDermott et al., 2004). We, therefore, established culture conditions to study the effect of insulin on the assembly of the droplets. We observed a strong influence of insulin on the process, and found that inhibition of PLD1 prevented the insulin-promoted stimulation. This indicates that PLD1 is a key player that connects insulin signal to the assembly of lipid droplets.
We also identified the kinase ERK2 as the previously described activator of lipid droplet formation in our cell-free system (Marchesan et al., 2003). Results from experiments in which ERK2 was overexpressed or microinjected into the cytosol, and from experiments in which ERK2 was inhibited by siRNA or the inhibitor Ste-Mek113, demonstrated that ERK2 also plays an important role in regulating the amount of cytosolic lipid droplets in intact cells. The effect of ERK2 on the formation of lipid droplets seemed to be very specific, since no effect of manipulating other MAP kinases (such as ERK1, JNK1 or JNK2, or p38 or p38ß) was observed.
The overexpression and knockdown experiments were based on the analyses of all cells in 20-50 randomly selected micrographs. Since the transfection efficiency was 50-70% the effect of this overexpression and knockdown of ERK2 and PLD1 is most probably underestimated. However, the observation that the effects of overexpression and microinjection of ERK2 are very similar, would argue against this being a major problem.
ERK2 also turned out to be of importance in the insulin-stimulated promotion of lipid droplet assembly. The observation that the formation of lipid droplets is strongly stimulated by insulin and that PLD1 and ERK2 are of central importance for this insulin signal is interesting, since the accumulation of cytosolic lipid droplets, particularly in liver and muscle, is intimately linked to the development of insulin resistance and its important complications, type 2 diabetes and cardiovascular disease (Browning and Horton, 2004; den Boer et al., 2004; Taskinen, 2003). Moreover, the accumulation of lipid droplets in macrophages (forming foam cells) is a landmark of both early and late atherosclerosis. Thus, these observations may add to our understanding of the role of cytosolic lipid accumulation in two of the most important metabolic disorders of the twenty-first century: insulin resistance and atherosclerosis.
The observation that both PLD1 and ERK2 are important for the effect of insulin on the formation of lipid droplets indicates that the two enzymes may influence each other's ability to stimulate the formation of these droplets. Indeed, we found that an inhibitor of ERK2 could eliminate the whole effect of PLD1 on the formation of the droplets. When the interactions between ERK2 and other proteins were investigated, we chose to use the inhibitor since this approach eliminated the problem of transfection efficiency. Ste-Mek113 had the same effect on lipid droplet formation as siRNA to ERK2.
We have ruled out the possibility that ERK2 is an activator of PLD1. In this respect, our results differ from the results obtained with the unpurified cytosolic activator in the cell-free system (Marchesan et al., 2003). This observation might be explained by the presence of a PLD1 activator [such as ARF or Rho family members, or protein kinase C (Hammond et al., 1997)] together with ERK2 in the cytosolic fractions. We also failed to detect any effect of PLD1 on the phosphorylation of ERK2 (supplementary material Fig. S8), arguing against an effect of PLD1 on the ERK2 activity in NIH 3T3 cells. Thus, our current model proposes that instead of activating PLD1, the cytosolic protein ERK2 appears to mediate the signaling pathway activated by PLD1, or possibly to function in parallel with it.
We observed that ERK2 influences the phosphorylation of two proteins that are of importance for the assembly of lipid droplets, i.e. dynein and caveolin. We have previously demonstrated that intact microtubules are essential for an important step in the assembly of lipid droplets, the fusion of droplets to form larger structures (Bostrom et al., 2005), and we could detect dynein on ADRP-containing lipid droplets (Bostrom et al., 2005). In agreement with the importance of dynein in the assembly of lipid droplets, we previously observed that vanadate, often used as an inhibitor of dynein, reduced the amount of cytosolic lipid droplets (Bostrom et al., 2005). However, vanadate is a rather unspecific inhibitor; thus, we addressed the question of the importance of dynein for the assembly of lipid droplets in a different way, by investigating the effect of microinjection of monoclonal antibodies to the intermediate chain of dynein. The results indicated that dynein plays an important role in the formation of lipid droplets.
It is well known that the activity of dynein is highly dependent on phosphorylation reactions (King, 2000; Kumar et al., 2000). Moreover, differential phosphorylation is one of the mechanisms behind the generation of the isoforms of the cytoplasmic dynein that are essential for the interaction between dynein and its correct cargo (King, 2000). This suggests that the ERK2-mediated phosphorylation of the dynein intermediate chain may be of importance for the function of the motor protein, including its ATPase activity (King, 2000; Kumar et al., 2000), or even for the interaction between dynein and the lipid droplet. The second possibility may find support in our observation that the dynein that is present on lipid droplets is phosphorylated. Moreover, we observed in cell-free experiments that ERK2 increased the amount of dynein that could be co-precipitated with ADRP. This is interesting, since it has been shown that dephosphorylation of dynein increases its association with the protein p150Glued, which mediates the association between dynein and organelles that are to be be transported (Vaughan et al., 2001). It is probably that the interaction between dynein and a structure surrounded by a membrane differs from that between dynein and a structure surrounded by a monolayer. However, the complexity of the phosphorylation of dynein should be kept in mind, and further studies are needed to clarify the exact relationship between the ERK2-mediated phosphorylation of dynein and the rate of assembly of cytosolic lipid droplets. Such studies must involve an elucidation of the role of ERK2 in the phosphorylation of dynein i.e. is it a direct phosphorylation or does ERK2 activate a second kinase? Moreover the phosphorylation site on dynein remains to be elucidated. It should be noted that both the light intermediate and the heavy chain contain the PX(S/T)P motif suggested to be an ERK2 phosphorylation site (Fantz et al., 2001). We believe however, that our results point to dynein as a possible mediator of the effect of ERK2 on the assembly of lipid droplets.
Based on the results discussed above, we propose a model for the role of PLD1 and ERK2 in the assembly of lipid droplets. The two enzymes, and in particular PLD1, are essential for the insulin-stimulated accumulation of cytosolic lipid droplets. PLD1 and ERK2 influence different parts of the mechanism that leads to the formation of lipid droplets. Thus, PLD1 provides the phosphatidic acid necessary for the formation of the droplets, whereas ERK2 phosphorylates dynein, a phosphorylation that is important for the correct interaction between dynein and its cargo (i.e. the lipid droplet) or the motor activity of dynein. Dynein has a central role in the assembly of lipid droplets though its influence on the fusion between droplets, but one might also speculate that the motor protein provides the energy needed to allow the droplet to bud from the microsomal membrane. This means that even if ERK2 does not influence the activity of PLD1, its action is a prerequisite for the influence of PLD1 on the assembly of lipid droplets.
|
Materials and Methods |
|---|
|
TopSummaryIntroductionResultsDiscussionMaterials and MethodsReferences |
|---|
Antibodies
Antibodies to ADRP were purchased from Research Diagnostics (Flanders, NJ) and antibodies to PLD1 were purchased from Biosource International (Camarillo, CA). Horseradish peroxidase-linked anti-guinea pig and anti-rabbit immunoglobulin (Ig) were from Dako (Glostrup, Denmark) and Amersham Biosciences (UK), respectively. Fluorescein-conjugated anti-guinea pig Ig and antibodies to ERK1, ERK2, dynein intermediate chain, vimentin, -tubulin and ß-actin were from Abcam (Cambridge, MA). Antibodies to GAPDH were from Ambion (Austin, TX) and antibodies to active Mapk were obtained from Promega (Madison, WI). Anti-JNK and anti-p38 were obtained from Cell Signaling. Anti-ß-COP was obtained from Oncogene Research Products. Anti-GRP78, anti-caveolin, anti-caveolin1, and phospho-specific antibodies to caveolin (pY14), were purchased from BD Biosciences Pharmingen. Anti-golgin97 was obtained from Molecular Probes.
Quantification of lipid droplets
Cells were fixed in 3.7% formaldehyde for 10 minutes, pretreated with 60% isopropanol for 30 seconds, and stained with Oil Red O in 60% isopropanol for 20 minutes. The cells were then washed with 60% isopropanol for 30 seconds, treated with hematoxylin for 20 minutes, and washed in cold water. The coverslips were mounted on microscope slides with Mowiol (Calbiochem, Darmstadt, Germany) and viewed with a Zeiss Axioplan 2 epifluorescence microscope. Images were obtained with an Axiocam camera and Axiovision 3.1 or 4.2 software, converted to JPEG format with the same software (resolution, 8 pixels/µm). Red pixels were identified by analyzing the color spectra (red, green, and blue). A pixel was considered red if it did not differ by more than a fixed value from the red spectrum (supplementary material Fig. S9A). Once all pixels were examined, the program identified adjacent red pixels as a lipid droplet `object' as described previously (Sonka et al., 1999).
The coefficient used to define the value for a red pixel (0.8) was established empirically by manually calculating the number of lipid droplets in 10 cells and comparing the values with those obtained by the computer program. Coefficients below 0.4 and above 0.9 gave nonsense values (1 and 0). The effect of variation between 0.4 and 0.9 is shown in supplementary material Fig. S9B. A coefficient of 0.8 resulted in almost complete agreement with the number of droplets counted manually. To confirm this, a new series of cells was stained, droplets were counted manually, and the results were compared with those obtained by the computer program using a coefficient value of 0.8. The results showed a very good correlation (r=0.9347) (see Fig. S9C in supplementary material).
In a second control experiment, we compared the droplet area determined by the program and by manual counting of pixels. There was complete agreement (r=0.9998). These results demonstrate that the program can be used to measure the number and total area of the Oil Red O-stained lipid droplets in the cell. Thus, the software allowed analysis of the total area of lipid droplets/cell. It identified all droplets and marked them, which made it possible to follow the process interactively.
The software was also extended to carry out three-dimensional reconstructions. This was done by integrating the planes obtained during confocal microscopy (20 slices of 0.3 µm per time point) using a previously described algorithm (Lorensen and Cline, 1998). Under these conditions, the pixel is either black or white, and the software uses the intensity to determine whether the pixel belongs to the droplet or not (supplementary material Fig. S9A). Adjacent pixels were identified and used to construct the three-dimensional image as described previously (Lorensen and Cline, 1998) and the volume was determined based on the thickness of the confocal planes. To verify that the software gave a correct volume estimation, we microinjected fluorescent beads (FluoSpheres® of known diameter and volume) into the cytosol of NIH 3T3 cells and analyzed the cells with confocal microscopy. The results (supplementary material Fig. S9D,E) showed an excellent correlation between the expected and determined volume (r=0.9997; P<0.001). It has been demonstrated (Fukumoto and Fujimoto, 2002) that staining with Oil Red O can change the structure of the droplets, owing to extraction of cellular phospholipids (DiDonato and Brasaemle, 2003). Since such a change may influence the size and volume of the droplets, we investigated whether there was any major difference in the size distribution between droplets stained with Oil Red O and with Nile Red or Bodipy. The results (supplementary material Fig. S9F) showed a very good correlation between the volumes determined under the different conditions (r=0.991; P<0.001). Thus, the volume of lipid droplets can be determined after staining with Oil Red O.
In the next experiment, we investigated whether the amount of lipid droplets could be estimated as the total area of Oil Red O-stained pixels in the cell. To do this, we incubated cells with oleic acid for different times and determined the total volume of Oil Red O-stained droplets/cell and the total area of the Oil Red O-stained droplets/cell in the same cells. The results (supplementary material Fig. S9G) showed that there was a good correlation between the two ways of determining the amount of lipid droplets in the cell (r=0.991; P<0.001).
As discussed above, the staining with Oil Red O can change the structure of the droplets, owing to extraction of cellular phospholipids. We therefore investigated whether there was any major difference in the size distribution between droplets stained with Oil Red O or with Nile Red as estimated by the 2D system. The results (supplementary material Fig S9H) revealed only small differences.
We also investigated the relationship between the area of Oil Red O-stained lipid droplets in the cell and the total amount of triglycerides (supplementary material Fig. S9I). NIH 3T3 cells were incubated with oleic acid for periods between 0 and 8 hours, and the quantity of cellular triglycerides was measured and normalized to cellular protein. In parallel experiments, the total area of Oil Red O-stained lipid droplets was measured at each time point in 20 randomly selected pictures of the Oil Red O-stained cells. The results (supplementary material Fig. S9I) demonstrated a highly significant linear relationship (r=0.93; P<0.001). Thus, there was a strong correlation between the total area of Oil Red O-stained lipid droplets and the triglyceride content of the cell.
Since we have also based conclusions on variations in total area of Oil Red O-stained lipid droplets below 500, we carried out a new experiment concentrating on this region of the curve. The cells were incubated with oleic acid for periods between 0 and 2 hours and the relationship between the total area of Oil Red O-stained lipid droplets and the amount of triglycerides per amount of cellular protein was followed. Also in this range of the curve, there was a linear relationship between the total area of Oil Red O-stained lipid droplets/cell and the amount of triglycerides per amount of cellular protein (r=0.89; P=0.015) (supplementary material Fig. S9J).
Identification of the cytosolic activator of lipid droplet formation
The fraction that activated the assembly of lipid droplets in the cell-free system was recovered from rat adipocytes by the chromatographic procedure described previously (the S-fraction) (Marchesan et al., 2003). The fraction was diluted 20-fold with 10 mM Tris, 1 M (NH4)2SO4, pH 7.0 (final volume 2 ml), and subjected to hydrophobic interaction chromatography on a Resource PheTM column using the Äkta Prime system (Amersham). The column was equilibrated with 10 mM Tris, 1 M (NH4)2SO4, pH 7.0, at a flow rate of 1.0 ml/minute. The elution was started with 25 ml of the same buffer and followed by a (NH4)2SO4 gradient (1 to 0 M in 10 mM Tris-HCl, pH 7.0; total volume 10 ml). The optical density (at 280 nm) and conductivity were recorded, and fractions were combined and analyzed for their ability to induce lipid droplet formation in the cell-free system as described previously (Marchesan et al., 2003). The fraction containing the major amount of activity was electrophoresed in a 10% SDS-polyacrylamide gel and stained with Sypro Ruby protein stain (Bio-Rad) as described by the manufacturer. Protein bands were cut out, trypsinized, cleaned and concentrated by chromatography on a ZipTip containing C18 material (Millipore), and analyzed by matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (MS) as described (Marchesan et al., 2003), except that Mascot software (available at www.matrixscience.com) was used for protein assignment. When the MALDI-TOF analysis was inconclusive, tandem MS/MS was performed with the quadrupole time-of-flight technique, and proteins were identified with Mascot software.
Cell culture
NIH 3T3 cells were cultured as recommended by the American Type Culture Collection. Incubation with oleic acid (360 µM in 0.3% BSA) was carried out as described previously (Boström et al., 1988).
SiRNA and plasmid transfections
Cells were transfected at 60% confluency by adding to each 10 cm2 culture well either 8 µl of siRNA (20 µM) and 4 µl of Lipofectamine 2000 in a total volume of 800 µl, or plasmids (4 µg DNA) and 10 µl of Lipofectamine 2000 in the absence of serum and antibiotics, as recommended by the manufacturer (Invitrogen). Transfection efficiency was 50-70%, as estimated from transfections with fluorescent siRNA or with a plasmid expressing green fluorescent protein (GFP). siRNAs used are listed in Table S1 in supplementary material. ß-actin, GAPDH and GRP 78 were used as control proteins. cDNA and plasmids used for transfection are listed in Table S2 in supplementary material.
Microinjection and time-lapse studies of living cells
This was carried out as described previously (Bostrom et al., 2005). The effect of the microinjected substances was investigated by examining the total area of Oil Red O-stained lipid droplets (see above) or fusion between droplets during time-lapse studies (Bostrom et al., 2005).
Other methods
SDS-PAGE, immunoblotting, and separation of radioactive triglycerides were carried out as described (Andersson et al., 1994). Western blots were quantified using MultiGauge software (Fuji Photo Film, Japan). Triglycerides were determined using a commercial kit (Roche/Hitachi, IN). The incorporation of radioactivity into phosphatidic acid (PA) was determined following a 24 hour incubation with [3H]palmitate (1 µCi/ml culture medium). The lipids were extracted from the cells using chloroform:methanol as described (Andersson et al., 1994) and the PA was isolated by two-dimensional thin layer chromatography as described (Steiner and Lester, 1972). Separation of phosphorylated and unphosphorylated proteins by the PhosphoProtein purification kit was carried out as recommended by the manufacturer. ß-oxidation of fatty acids was determined as described (Hansson et al., 2004). The PLD activity was determined as described previously (Marchesan et al., 2003). The Mann-Whitney rank-sum test, Student's t-test or one-way ANOVA were used for statistical analysis, as indicated in the text.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
* These authors contributed equally to this work 
|
References |
|---|
|
TopSummaryIntroductionResultsDiscussionMaterials and MethodsReferences |
|---|
Andersson, M., Wettesten, M., Borén, J., Magnusson, A., Sjöberg, A., Rustaeus, S. and Olofsson, S.-O. (1994). Purification of diacylglycerol:acyltransferase from rat liver to near homogeneity. J. Lipid Res. 35, 535-545.[Medline]
Barr, R. K., Kendrick, T. S. and Bogoyevitch, M. A. (2002). Identification of the critical features of a small peptide inhibitor of JNK activity. J. Biol. Chem. 277, 10987-10997.
Bonny, C., Oberson, A., Negri, S., Sauser, C. and Schorderet, D. F. (2001). Cell-permeable peptide inhibitors of JNK: novel blockers of beta-cell death. Diabetes 50, 77-82.
Boström, K., Borén, J., Wettesten, M., Sjöberg, A., Bondjers, G., Wiklund, O., Carlsson, P. and Olofsson, S.-O. (1988). Studies on the assembly of apo B-100-containing lipoproteins in HepG2 cells. J. Biol. Chem. 263, 4434-4442.
Bostrom, P., Rutberg, M., Ericsson, J., Holmdahl, P., Andersson, L., Frohman, M. A., Boren, J. and Olofsson, S. O. (2005). Cytosolic lipid droplets increase in size by microtubule-dependent complex formation. Arterioscler. Thromb.
Not significant (NS) K898R PLD1 vs empty vector, one-way ANOVA). (C) Accumulation of radioactive triglycerides after incubation with [3H]palmitic acid for the indicated times in NIH 3T3 cells transiently transfected with wild-type PLD1 (triangles) or K898R PLD1 (squares). Values are mean ± s.d. from cells from three different culture dishes. (D) Influence of the transfection with PLD1 on the levels of ADRP in the cell. Cells were transfected as in A and the cell lysate was blotted against antibodies to ADRP. (E) Effect of transient transfection of NIH 3T3 cells with wild-type PLD2 (PLD2 wt), K758R PLD2 (a catalytically inactive mutant) or an empty vector on the area of Oil Red O-stained lipid droplets per cell (values are mean ± s.d. of all cells in 20 randomly selected micrographs from each group; NS). (F) The amount of ADRP as estimated by immunoblotting.
