|
|
||||||||
THEMES
1Department of Molecular and Cellular Biology and 2Division of Diabetes, Endocrinology, and Metabolism, Department of Medicine, Baylor College of Medicine, Houston, Texas
Submitted 12 December 2006 ; accepted in final form 18 December 2006
| ABSTRACT |
|---|
|
|
|---|
adipose differentiation-related protein; fatty liver; obesity
Adipose differentiation-related protein (ADRP), the major LD protein, was first cloned by Ginette Serrero and her research group in an attempt to find genes upregulated during adipose differentiation, using a mouse adipogenic cell line (8). Originally thought to be an adipose-specific protein, ADRP was later found to be ubiquitously expressed (1, 7). The gene symbol was changed to Adfp in 1993; Heid et al. (7) proposed the name adipophilin (or ADPH) for the human homolog.
The human ADFP gene is located on chromosome 9 and the mouse homolog on chromosome 4. A nonprocessed pseudogene was found in mouse on the same chromosome
23 kb from the functional gene, probably a result of gene duplication. The pseudogene seems not to be expressed by RNase protection assay. Using gene-specific primers and RT-PCR, we were also unable to detect the potential transcript from mouse liver.
PAT Domain Proteins and Lipid Droplet Targeting
ADFP shares sequence similarity with other genes collectively known as the PAT domain-containing gene family, a term first coined by Lu et al. (10) to highlight the three major genes in this family, perilipin, Adfp, and Tip47. Other members of this family include MLDP (also known as Pat-1 or OXPAT) in vertebrates and lipid storage droplet (LSD) proteins Lsd1 and Lsd2 in insects. The
100 residues at the NH2 terminus of these proteins share substantial sequence similarity, ranging from 40% between PLIN and ADFP to 65% between TIP47 and ADFP, and are called the PAT-1 domain; PAT-2 domain is assigned to the region toward the COOH-terminal portion, which is not as well conserved (20% identity between PLIN and ADFP and 40% between TIP47 and ADFP). A string of 29 tandem repeats of a 33-amino acid motif is present in yet another protein called S3-12. The 33-amino acid motif is present in PAT protein family as a single copy in the PAT-2 domain; it displays
50% identity between ADFP and TIP47 but only 27% between PLIN and MLDP.
Although these domains are conserved across paralogs, their functional significance is unclear. Other than PLIN, which is involved in the regulation of lipolysis, no known function has been unequivocally attributed to these proteins. The presumptive function for PAT domain is LD targeting, since multiple proteins associated with LD share this domain. However, neither the PAT-1 nor PAT-2 domain is absolutely required for LD targeting (13). The functional unit of ADFP that mediates LD targeting seems to be discontinuous and redundant, whereas other LD proteins such as PLIN and TIP47 seem to target LD through hydrophobic regions. Recently, studies on the Parkinson's disease-related protein
-synuclein identified the presence of 11-amino acid repeats found in classic soluble (exchangeable) apolipoproteins; such repeats are known to form amphipathic
-helices that mediate protein-lipid binding (2). A similar 11-mer repeat is also found in ADFP and the PAT domain-containing proteins, suggesting that it may contribute to LD targeting in these proteins.
Intracellular Localization of ADFP
ADFP is found on the surface of the LDs in all the cells examined (1, 7). It was initially reported to be localized in the plasma membrane and microsome and not cytosol (8), but subsequent studies cast doubt on these latter locations (13). Recently, using freeze-fracture scanning electron microscopy (SEM), Robenek et al. (19) localized ADFP to the plasma membrane. SEM also showed that ADFP is localized not only on the surface but also inside the core of the LD (19). The different conclusions on ADFP subcellular localization may be a reflection of the sensitivity of the detecting method, the cell type, and the culture conditions in cell studies. Since ADFP does not have a predicted membrane-spanning motif, the presence of this cytosolic protein in plasma and intracellular membranes and in LD may also be mediated by posttranslational lipid modification. There are eight to nine myristoylation consensus sites present in the ADFP sequence. In addition, although ADFP could also interact with other LD proteins via a leucine zipper motif present at the COOH terminus of ADFP, deletion of this motif does not affect LD targeting (13).
Regulation of Adfp Expression
Long-chain free fatty acids (FFAs) induce the expression of Adfp in different cell lines (4, 6); polyunsaturated fatty acids are particularly effective in this respect. Fasting, which stimulates lipolysis and raises serum FFAs, stimulates Adfp mRNA and protein expression in the mouse liver (9). Treatment of rats with a CPT-1 inhibitor, etomoxier, blocks mitochondrial
-oxidation, resulting in the upregulation of Adfp mRNA probably through an increased intracellular FFA pool. FFAs are ligands for the peroxisome proliferator-activated receptors (PPARs), nuclear hormone receptors that are important transcription factors regulating cellular lipid homeosis, proliferation, and inflammatory responses. PPAR, upon ligand activation, dimerizes with another nuclear receptor, RXR (retinoid X receptor), and binds to a specific region of the promoter, a cis element called PPRE, of its target genes. There are three major subtypes of PPARs,
,
, and
, each encoded by a separate gene. A PPRE is present in the promoter region
2 kb upstream from the transcription initiation site of the Adfp gene and binds to all three PPAR subtypes in mobility shift assays in vitro (4). Consistent with this, Adfp mRNA can be upregulated by all three PPAR agonists in vitro and in vivo (4, 5). In addition, WY-14643, a synthetic PPAR-
agonist, induces Adfp mRNA expression in wild-type mice but fails to do so in the PPAR-
knockout mice (4), supporting the involvement of PPARs in the regulation of Adfp gene expression.
As stated earlier, the upregulation of Adfp mRNA occurs in conditions in which the FFA flux is high, such as fasting in vivo or FFA incubation in vitro. In contrast, de novo FFA synthesis does not affect Adfp mRNA expression (4). It is tempting to hypothesize that hepatocytes are able to distinguish the endogenous versus exogenous sources of FFAs and respond with appropriate gene regulation. Hence, under surplus conditions when exogenous FFAs are abundant, hepatocytes will use a mechanism involving ADFP for storing these FFAs as triglyceride (TG) for future use, whereas, under conditions when exogenous FFAs are limiting, they will initiate de novo fatty acid synthesis and use these newly synthesized FFAs to make TG for VLDL secretion to provide TG to peripheral tissues. Both processes require TG synthesis. Since there are two separate genes regulating the final step of TG synthesis, namely, diacyl glycerol acyl transferases DGAT1 and DGAT2, it is possible that each is responsible for TG synthesis for a particular compartment in the cell (but see below for further discussion).
In addition to PPARs, Adfp gene transcription is also regulated by the transcription factors PU.1 and Ap-1 in macrophages (23). These transcription factors bind to Est/Ap-1 composite element present in the Adfp promoter in response to phorbol 12-myristate 13-acetate (PMA), a protein kinase C (PKC) activator. Adfp mRNA can also be regulated by gonadotropin and prostaglandin in primate periovulatory follicles (20).
At the protein level, ADFP is regulated by ubiquitin-mediated proteasome degradation pathway. Analogous to what was found in another lipid-binding protein, apoB100, which is produced in abundance but is mostly degraded intracellularly unless it is fully lipidated in the endoplasmic reticulum, ADFP is degraded unless it is "protected" by LDs (25), whose formation in the cytosol is stimulated by incubating cells in oleic acid and in animals by fasting (resulting in de novo lipogenesis) or high-fat diet feeding.
LD Formation
The conventional view of LD formation postulates that a "pouch" is generated in the outer leaflet of the endoplasmic reticulum (ER) membrane where TG is synthesized. This distension pinches off from the ER to form a primordial LD in the cytosol. Proteins present in LDs either are integral ER membrane proteins or are added to the growing LD from the cytosol. The putative role of ADFP in LD formation using this model was reviewed recently (24). It is noteworthy, however, that an alternative model based on data obtained from freeze-fracture electron microscopy and without the need for a primordial LD peeling off the ER outer member has been advanced by Robenek et al. (18).
Function of ADFP
So far, no known function has been definitively attributed to ADFP. Although in vitro studies have suggested a possible involvement of ADFP in fatty acid absorption and transport, our in vivo study does not support this notion (3). In the following paragraphs we discuss the potential function of ADFP in different tissues.
Adipose tissue. Although ADFP is highly upregulated during adipocyte differentiation, we found that adipocyte differentiation is not affected by the absence of ADFP in mice in vivo. Furthermore, adipose lipolysis rate is preserved in both basal and stimulated conditions in the absence of ADFP. We detected no compensatory increase in TIP47 or PLIN for the loss of ADFP in chow-fed Adfp knockout mice. Despite the early increase in Adfp mRNA during adipocyte differentiation and its continuous presence throughout the differentiation process, ADFP protein is downregulated when PLPIN protein appears during adipocyte differentiation, and eventually, at the protein level, PLIN completely replaces ADFP in mature adipocytes. Although ADFP is upregulated in the fat of Plin-knockout mice, it cannot substitute for the function of PLIN, and Plin-knockout mice have high basal lipolysis and extreme reduction in adipose tissue mass. These and other effects of the absence of PLIN render these animals resistant to genetic and high-fat diet-induced obesity (11). Thus, although ADFP appears before other LD proteins during adipocyte differentiation, it has not been assigned any dominant function in adipocytes. However, we cannot exclude its possible involvement in physiological or pathological perturbations such as long-term high-fat diet feeding or fasting.
Liver. Generation of Adfp knockout (Adfp/) mice has uncovered an unexpected function of ADFP in mice (3). Adfp knockout mice fed regular chow diet display a 60% reduction in hepatic TG compared with the wild-type mouse, whereas other hepatic lipids remain unaffected. Although hepatic TG is reduced, the fatty acid composition of the TG is unchanged, i.e., the relative percentage of each fatty acid species in hepatic TG is the same between wild-type and knockout mice. FFA uptake by hepatocytes isolated from Adfp/ mice is normal. These findings suggest that ADFP is probably not involved in the transport of FFAs as was suggested by in vitro studies. Other experiments have shown that the reduction in hepatic TG in Adfp deficiency cannot be explained by a change in lipid uptake, synthesis, utilization, or secretion.
Because VLDL is rich in TG, the finding of normal VLDL secretion in an Adfp-deficient liver that has only 40% normal TG content was counterintuitive. On the other hand, hepatic TG concentration does not always correlate with the level of VLDL output, e.g., leptin-deficient (ob/ob) mice have very high levels of hepatic TG, but their VLDL secretion is impaired. To determine whether intracellular lipid partitioning might have been affected in Adfp-deficient liver, we examined the lipid profiles in hepatocyte cytosolic, microsomal, and nuclear fractions. We found that TG is reduced in all compartments of the Adfp knockout liver compared with their wild-type counterparts with the exception of the microsomal fraction, which was paradoxically higher in knockout than in wild-type mice. TG is synthesized on the membranes of the ER, where the diacyl glycerol acyl transferases DGAT1 and DGAT2 are located. It is unclear in which compartment the two enzymes are located and how the newly synthesized TG are distributed, because recent studies using adenovirus overexpression of DGAT1 and DGAT2 in vivo generated conflicting results (15). Irrespective of where the TG are synthesized (in the ER lumen or in cytosol), they can be sequestered in the cytosolic LDs through a process yet to be defined in which ADFP plays a crucial role; alternatively, they are transported to the ER lumen for the assembly and secretion of apoB-containing VLDL via the obligatory action of microsomal triglyceride transfer protein (MTP). We did not find a change in hepatic DGAT activity (3) or an alteration of DGAT1 or DGAT2 expression (unpublished observations) in the Adfp/ liver, making it unlikely that they mediate the increase of TG in the microsome. We did find, however, that MTP protein level is increased in the Adfp knockout liver compared with wild type. We believe that it is the upregulation of MTP that causes the accumulation of TG in the ER compartment in the absence of ADFP. MTP upregulation also accounts for the normal VLDL secretion in the presence of a reduction of hepatic TG content in the liver of these animals. It is a plausible mechanism for the overall reduction of TG in the knockout livers.
Liver-specific MTP inactivation in mice abrogates VLDL secretion and causes TG accumulation in the hepatocytes, mainly in the cytosol. In these animals, the hepatic ADFP and TIP47 proteins are all upregulated (unpublished observations). Recent studies showed that overexpression of ADFP either by in vivo delivery of recombinant adenovirus or by induction with PPAR-
agonist causes an increase of cytosolic TG by reducing VLDL secretion (5). These findings indicate that ADFP is capable of sequestering TG in the cytosol, diverting it from entering into ER lumen for VLDL secretion.
Thus across the opposite sides of the ER membranes are two proteins, ADFP in the cytosol and MTP in the ER lumen, that exercise opposite forces in retaining or transporting TG to its specific compartment. These two proteins, perhaps together with DGAT1 and DGAT2, determine the balance between whether TG should be stored in the cytosolic LDs or secreted as VLDL from the liver.
In addition to TG reduction, it was also noted that the number of LDs is reduced in the Adfp knockout hepatocytes when the mice are fed high-fat diet for 4 wk. In the normal liver, LD is not readily observable under low-power light microscopy, but it becomes visible after high-fat diet. Not only is the LD number reduced, but the size distribution of the LDs is also changed with marked reduction in numbers for the largest and the smallest. These findings further support suggestions that ADFP is involved in LD formation and/or maturation.
Macrophages. Macrophages play an important role in atherosclerosis. An early event in atherosclerosis is the accumulation of LDs in lesional macrophages associated with ADFP accumulation. ADFP is the most abundant LD-associated protein found in these cells. Modified lipoproteins, e.g., oxidized LDL or acetylated LDL, that are highly atherogenic upregulate Adfp expression in macrophages in vitro (22). Furthermore, Adfp mRNA is upregulated in human atherosclerotic plaques compared with lesion-free areas of the same arteries (9). Conversely, ADFP overexpression in THP-1 macrophage enhances lipid accumulation and prevents lipid efflux. These results suggest that ADFP is potentially a proatherosclerogenic protein. Our preliminary study using Adfp-deficient mice seems to support this notion.
Lung. Although it has been proposed that ADFP may be involved in the transfer of lipids from LDs to the type 2 lung epithelial cells for the production of surfactant phospholipids, such a role has not been substantiated (14).
Mammary gland. LDs are the essential components of milk that provide an energy source and essential fatty acids for membrane synthesis in newborns. Mammary epithelial cells release LDs into the lumen of mammary alveoli by exocytosis, creating a LD with a globule structure wrapped in three layers of phospholipids known as milk fat globules. ADFP is the major LD protein in milk. Two other proteins, the membrane-bound butyrophilin and the cytosolic enzyme xanthine oxidoreductase, may be involved in attaching the secretory granules to the plasma membrane through their interaction with ADFP at the milk secretory granule surface (12). Using data obtained by freeze-fracture immunocytochemistry, Robenek et al. (17) recently proposed a new model for milk fat globule formation and secretion. They found that the physical localizations of butyrophilin, xanthine oxidoreductase, and ADFP are incompatible with a complex formation among these three proteins in the secretory granules (17) and suggested that butyrophilin alone controls the globule formation and secretion. Further studies using butyrophilin-null mice might provide further insight on the process.
Muscle. ADFP is the predominant LD-associated protein in skeletal muscle in humans (16). Furthermore, muscle ADFP is lower in insulin-resistant subjects, a situation that can be reversed by weight reduction or by troglitazone treatment coincident with an improvement in glucose tolerance. It is possible that the upregulation of ADFP may help sequester fatty acids as TG in LDs, protecting the muscle from the detrimental effects of fatty acids on insulin action and glucose homeostasis (16).
Conclusions
ADFP has long been recognized as a universal marker for intracellular LD in tissues, but its function is just beginning to be elucidated. The phenotypes of Plin and Adfp knockout mouse models indicate that ADFP plays a minimal role in adipose tissue homeostasis. It is, however, a major determinant of TG content in the liver. ADFP and MTP seem to play a yin-yang role in TG homeostasis in the liver as the two proteins sequester the intracellular TG in separate compartments for storage and secretion, respectively. The role of ADFP in foam cell formation and atherosclerosis is under investigation. Preliminary results suggest that ablation of Adfp reduces atherosclerotic lesion development. The function of ADFP in muscle, adrenal gland, mammary gland, and other tissues requires further investigation. Although Adfp-null mice are a powerful tool for dissecting the role of Adfp in different organs and tissues, we must be mindful of the possible functional compensation by TIP47 and other LD proteins for the loss of ADFP (25). The generation of single or multiple genetic knockouts or knockdowns (e.g., using RNA interference) of LD proteins will continue to advance our understanding of LD biology and function and its application toward the prevention and treatment of obesity, fatty liver, and atherosclerosis.
| GRANTS |
|---|
|
|
|---|
| FOOTNOTES |
|---|
| REFERENCES |
|---|
|
|
|---|
-synuclein and other exchangeable lipid binding proteins. J Mol Biol 329: 763778, 2003.[CrossRef][ISI][Medline]
activators and fasting induce the expression of adipose differentiation-related protein in liver. J Lipid Res 47: 931943, 2006.
activation increases triglyceride mass and adipose differentiation-related protein in hepatocytes. J Lipid Res 47: 329340, 2006.This article has been cited by other articles:
![]() |
O. Beaslas, F. Torreilles, P. Casellas, D. Simon, G. Fabre, M. Lacasa, F. Delers, J. Chambaz, M. Rousset, and V. Carriere Transcriptome response of enterocytes to dietary lipids: impact on cell architecture, signaling, and metabolism genes Am J Physiol Gastrointest Liver Physiol, November 1, 2008; 295(5): G942 - G952. [Abstract] [Full Text] [PDF] |
||||
![]() |
J. Yan, B. M. Barnes, F. Kohl, and T. G. Marr Modulation of gene expression in hibernating arctic ground squirrels Physiol Genomics, January 17, 2008; 32(2): 170 - 181. [Abstract] [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |