U18666A

Overexpression of A-kinase anchoring protein 12A activates sterol regulatory element binding protein-2 and enhances cholesterol efflux in hepatic cells

A-kinase anchoring protein 12 (AKAP12) is known to function as a scaffold protein and as a putative tumor suppressor. However, little is known about the biological role of AKAP12 in hepatic cells. In this study, we performed micro-array analysis to identify the down- stream pathway of AKAP12A, and found that AKAP12A overexpression up-regulates the expressions of several cholesterol-associated genes including HMG-CoA reductase and LDL receptor, which have been reported to be controlled by sterol regulatory element bind- ing protein-2 (SREBP-2). It was found that AKAP12A activates SREBP-2 in hepatic cells, as demonstrated by the presence of its cleavage product, whereas the activation of sterol regulatory element binding protein-1 was not remarkably changed. Moreover, AKAP12A- induced SREBP-2 activation was found to depend on SREBP cleavage-activating protein (SCAP), as inhibition of SCAP by RNAi or sterols blocked SREBP-2 activation in response to AKAP12A overexpression. Interestingly, the hydrophobic amine U18666A caused dramatic movement of AKAP12A from the plasma membrane to cytosol and lysosomal membranes. Moreover, cholesterol depletion from the plasma membrane (using methyl-β-cyclodextrin) caused a shift of AKAP12A from the plasma membrane to the cytoplasm. Cholesterol bind- ing assay revealed that the N-terminal region of AKAP12A binds directly to cholesterol in vitro. Furthermore, AKAP12A overexpression enhanced [3 H]-cholesterol efflux to extracel- lular acceptors, suggesting that AKAP12A may activate SREBP-2 by increasing cholesterol efflux. In conclusion, the present study suggests that AKAP12A is a novel regulator of cellular cholesterol metabolism.

1. Introduction

A-kinase anchoring proteins (AKAPs) bind the RII subunits of cAMP-dependent kinase (PKA), which cause a distinct subcellular localization of PKA within cells (Dell’Acqua and Scott, 1997; Wong and Scott, 2004). A- kinase anchoring protein 12 (AKAP12) is the official gene symbol of a group of orthologous proteins that include human gravin and the rodent Src-Suppressed C Kinase Substrate (SSeCKS). Gravin acts as a kinase scaffold protein associated with PKA and PKC (Nauert et al., 1997). Gravin also binds β2-adrenergic receptor (Fan et al., 2001), and the suppression of gravin leads to the disruption of recep- tor recycling in A431 cells (Lin et al., 2000a). Moreover, it has also been reported that gravin and SSeCKS have tumor suppressing activity (Choi et al., 2004; Lin et al., 2000b; Xia et al., 2001; Yoon et al., 2007). We previously reported that the expression of AKAP12 is down-regulated by aber- rant DNA methylation in gastric carcinoma, and that the re-expression of AKAP12A in AKAP12-nonexpressing AGS cells inhibits cell growth (Choi et al., 2004). In this study, we performed micro-array analysis using AGS cells to iden- tify the AKAP12A-regulated pathway. Initially, we hoped to find a cancer-associated pathway, but in the event obtained unexpected findings related to cholesterol metabolism.

Cholesterol is an essential lipid in mammalian cells and plays an important role in several cellular functions, such as membrane trafficking and signal transduction, and thus, cellular cholesterol levels are tightly regulated. The biosynthesis of cholesterol and other lipids is regulated by sterol regulatory element binding proteins (SREBPs) (Brown and Goldstein, 1997; Goldstein et al., 2006; Rawson, 2003). SREBPs consist of three transcription factors, SREBP-1a, SREBP-1c and SREBP-2. SREBP-1 prefer- entially regulates genes involved in fatty acid biosynthesis, whereas SREBP-2 mainly mediates cholesterol biosynthesis (Horton et al., 2002). They are synthesized as precursor pro- teins in the endoplasmic reticulum (ER), where they form a complex with SREBP cleavage-activating protein (SCAP) (Sakai et al., 1997). SCAP serves as sterol sensor and SREBP escorting protein. When intracellular cholesterol concen- trations are low, SCAP escorts SREBPs from the ER to the Golgi complex (Brown and Goldstein, 1999; Hua et al., 1996; Nohturfft et al., 2000; Radhakrishnan et al., 2004) where they are cleaved by Site 1 and Site 2 proteases (DeBose- Boyd et al., 1999; Zelenski et al., 1999), which causes the nuclear translocation of active transcription factor to direct the transcriptional activation of genes containing sterol response element (SRE) in their promoter regions (Magana and Osborne, 1996). On the other hand, when cells are over- loaded with cholesterol, SCAP binds cholesterol, undergoes a conformational change (Brown et al., 2002), and inter- acts with the ER retention proteins, INSIG-1 (Yang et al., 2002) and -2 (Yabe et al., 2002). As a result, SREBP cannot be transferred to the Golgi complex, and the biosynthe- sis of cholesterol stops because of decreased transcription of enzymes involved in the cholesterol biosynthesis path- way. SREBPs induce new SREBPs synthesis by way of SREs located in the promoter regions of their genes (Sato et al., 1996). SREBP-2 also regulates the expression of LDL receptor (Hua et al., 1993), which enables hepatocytes to remove cholesterol contained in LDL particles from the bloodstream.

Mammalian cells uptake cholesterol mainly from low-density lipoprotein (LDL), via the LDL receptor pathway (Brown and Goldstein, 1986). LDL binds to LDL receptors located at the plasma membrane, and subsequently endo- cytosed LDL is delivered to late endosomes, where it is hydrolyzed thereby releasing free cholesterol. This choles- terol is then transported rapidly out of endosomes to the plasma membrane before it moves back to the ER. Cellular cholesterol efflux is a crucial event during cholesterol trans- port, and enhancing efflux from liver cells increases plasma HDL concentrations (Maxfield and Tabas, 2005; Yokoyama, 2005). Several proteins play an important role in this pro- cess (Chang et al., 2006). For example, caveolin-1, a major protein marker of caveolae, enhances cholesterol efflux in HepG2 cells (Fu et al., 2004). However, in contrast to the well-studied role of cholesterol biosynthesis in cells, the cholesterol transport pathway is partially understood.

Here, we describe the effect of inducible AKAP12A overexpression on cellular cholesterol metabolism in hep- atic cells. The prominent phenotype is elevated HMG-CoA reductase (HMGCR) and LDL receptor expression via SREBP- 2 activation. We also provide evidence that AKAP12A binds cholesterol and enhances cholesterol efflux, phenomena that are probably interconnected.

2. Materials and methods

2.1. Reagents

25-Hydroxycholesterol (25-HC), methyl-β-cyclo- dextrin-conjugated cholesterol, methyl-β-cyclodextrin (MCD), Filipin, U18666A, and Tubulin antibody were from Sigma; SREBP-1 antibody, and SREBP-2 antibody from BD; LAMP-2 antibody from Santa Cruz; HMG-CoA reductase antibody, and Protein G agarose from Upstate; Anti-his antibody from Amersham; ApoAI, HDL, and z-VAD-fmk from Calbiochem; Ni-NTA agarose bead from Qiagen; [3H]-cholesterol from PerkinElmer; Lysotracker-red, Alexa Fluor 488 goat anti-rabbit IgG, and Alexa Fluor 594 goat anti-mouse IgG from Molecular Probes; SiRNAs from PROLIGO; and Amicon Ultra-15 30K was from Millipore. Polyclonal anti-AKAP12 antibody was used, as previously described (Choi et al., 2004).

2.2. Plasmid constructs

GFP-Akap12α was used, as previously described (Streb et al., 2004). To produce His-fusion proteins, AKAP12A cDNA was first cloned into pENTRY4 (Invitrogen). pENTRY4-AKAP12A deletion plasmids were constructed using either restriction enzymes (1–1782, 1–913, 915–1782, 1–640, and 1–181) or by amplifying the indicated regions by PCR (1–294, 1–165, and 552–943). To construct 17–1782 mutant in which N-myristoylation site is deleted, primers were designed to delete amino acids 1–16. The glycine- to-alanine substitution mutant G2A of pENTRY4-AKAP12A was engineered by standard protocol for PCR-mediated mutagenesis. pENTRT3C-Hpc2 was constructed by PCR (Kagey et al., 2003). All of the polymerase chain reac- tions were performed using Pfu polymerase (Stratagene). All constructs were confirmed by DNA sequencing. These constructs were cloned into pDEST17 Gateway vector (Invitrogen) to express His-fusion proteins.

2.3. Cell culture and adenovirus infection

All cell types used were grown at 37 ◦C in a 5% CO2 atmosphere. Human gastric cancer cells (AGS), and human hepatocellular carcinoma cell lines, SNU-354, —368, —387, —423, —449, —739, Hep3B, and HepG2 (Korean Cell Line Bank, Seoul) were cultured in RPMI 1640 supplemented with 10% FBS (fetal bovine serum) and gentamicin (10 µg/ml). COS-7 cell lines (American Type Culture Collec- tion) were maintained in DMEM supplemented with 10% FBS. The following adenoviral vectors have been described previously (Choi et al., 2004): Ad-empty, Ad-lacZ, and Ad- AKAP12A. Cells were seeded, and infected separately with these adenoviral vectors in 10% FBS/RPMI, and then incu- bated at 37◦ for 2 h with brief agitation every 15 min. The culture medium was then replaced with normal cul- ture medium, and the infected cells were returned to a 37◦ incubator. Cells were treated with z-VAD-fmk (60 µM) or 25-HC (1 µg/ml) and/or cholesterol (10 µg/ml) for 24 h when medium was changed.

2.4. Northern blot, RT-PCR and Western blot analysis

Northern blot, RT-PCR and Western blot analysis were performed as described previously (Choi et al., 2004; Lee et al., 2006). The primer sequences used to prepare cDNA probes and RT-PCR primers were as follows: MVD, 5r-AGCGCTGAGAAGAAGCTGAC-3r and 5r- GTGTCGTCCAGGGTGAAGAT-3r; HMGCR, 5r-GGCATTTGAC- AGCACTAGCA-3r and 5r-CTTTGCATGCTCCTTGAACA-3r; SREBP-2, 5r-AAGTCTGGCGTTCTGAGGAA-3r and 5r-CACAA- AGACGCTCAGGACAA-3r; SCAP, 5r-CCCCTACCTTGTGGTGG- TTAT-3r and 5r-GGACAGGACAGTGGTGAAAAA-3r; LDLR, 5r- CTCGCTGGTGACTGAAAACA-3r and 5r-CAAAGGAAGACGA- GGAGCAC-3r; INSIG1, 5r-GTGTTCCTGAAAAGCCCCATAGTG- 3r and 5r-GAAATATGGCAACTTGTTTTAATCAC-3r.

2.5. RNA interference

SNU-449 cells were seeded into 6 well plates. After 16–24 h, 50 nM of siRNA oligonucleotides specific for AKAP12A or an unspecific control oligonucleotide were transfected using Lipofectamine 2000. To deplete SCAP, SNU-449 cells were transfected with unspe- cific control or SCAP siRNA. After 24 h, cells were infected with Ad-empty or Ad-AKAP12A, and then further incubated for 24 h. The following sequences were used: unspecific control (sense, 5r-UUCUCCGAAC- GUGUCACGUtt-3r, antisense, 5r-ACGUGACACGUUCGGAG- AAtt-3r); AKAP12A (sense, 5r-CACCAUCAAUGGCGUAGCUtt-3r, antisense, 5r-AGCUACGCCAUUGAUGGUGtt-3r); and SCAP (sense, 5r-GCAUCAAGUUCUACUCCAUtt-3r, antisense, 5r-AUGGAGUAGAACUUGAUGCtt-3).

2.6. Immunofluorescence and confocal microscopy

To detect endogenous AKAP12, SNU-449 cells were fixed in 3.7% formaldehyde in PBS and permeabilized with 0.1% Triton X-100. Cells were then blocked with 1% bovine serum albumin (BSA) in PBS and incubated with anti-AKAP12 antibody, followed by Alexa Fluor 488 goat anti-rabbit IgG. Cholesterol was visualized by incubating coverslips in 50 µg/ml filipin for 1 h. Immunofluorescence images were obtained using a microscope (model BX51; OLYM- PUS). To detect exogenous Akap12α, COS-7 cells seeded on coverslips were transfected with GFP-Akap12α. Cells were incubated in DMEM medium containing 100 nM Lyso- tracker Red for 1 h prior to 3.7% formaldehyde fixation. Cells were imaged under a confocal fluorescence microscope (model LSM5; Carl Zeiss).

2.7. His-fusion proteins and cholesterol binding assay

His-fusion proteins were expressed in Escherichia coli BL21 strain by induction with 1 mM IPTG for 1 h. Fusion proteins were purified from crude lysates by affinity chro- matography on Ni-NTA agarose beads (Qiagen), according to the manufacturer’s instructions. After buffer exchange (Buffer A: 50 mM Tris–Cl, pH 8.0, 300 mM NaCl) using centrifuge concentrator Amicon Ultra-15, protein amounts were determined using BCA assays. The integrities of the His-fusion proteins were confirmed on SDS-PAGE by Coomassie blue staining. Cholesterol binding assays were performed as previously described (Ko et al., 2003; Radhakrishnan et al., 2004) with some modification. Briefly, each reaction was performed in a final volume of 100 µl of Buffer A, containing 100 nM of [3H]-cholesterol and 5 µg of His-fusion proteins. After incubation for 1 h at room temperature, mixtures were diluted by adding 0.4 ml of Buffer A, and then loaded onto columns packed with Ni-NTA agarose beads. Columns were washed 8 times with 1 ml Buffer A and then eluted with 1 ml Buffer A containing 200 mM imidazole. Eluates were assayed for radioactivity using a liquid scintillation counter. Peptides corresponding to the three PCDs (PCD1: 170–189, PCD2: 297–317, PCD3: 510–536) and the control (150–169) were synthesized with additional 6 × His at N-terminus by Peptron (Daejon, Korea). The four peptides were dissolved in DMSO, because PCD1 is hydrophobic and insoluble in water. Reactions were performed in Buffer A containing 100 nM of [3H]-cholesterol and 20 µg of each peptide over 1 h at room temperature. At the same time, 20 µg of anti-his anti- body was incubated with Protein G immobilized on agarose for 1 h at room temperature, and then anti-his conjugated beads were added to the [3H]-cholesterol–peptide com- plex. After further incubation for 1 h at room temperature, mixtures were washed 5 times with Buffer A and then eluted with 0.5 M ammonium acetate (pH 3.0). Eluates were assayed for radioactivity using a liquid scintillation counter.

2.8. Analysis of cholesterol efflux

SNU-449 cells were seeded on 12-well plates and cul- tured in a medium containing 10% FBS. To label cellular cholesterol, cells were incubated in 10% serum-containing medium with [3H]-cholesterol (1 µCi/ml) for 48 h, and infected with adenoviral vectors for the last 24 h. After labeling, cells were washed 3 times with PBS, and 1 ml of serum-free medium was added per well. Efflux stud- ies were then carried out using BSA (2 mg/ml), BSA plus ApoAI (20 µg/ml), HDL (50 µg/ml) or 10% serum as accep- tors. After 4 h, the media were harvested and centrifuged, and the radioactivity in the supernatant and in the cells was measured. Cholesterol efflux is expressed as the per- centage of labeled cholesterol transferred from cells to the medium.

3. Results

3.1. AKAP12A overexpression increased the mRNA levels of enzymes involved in cholesterol biosynthesis and activated SREBP-2 in AGS gastric cancer cells

Total RNA from AGS cells infected with Ad-empty or Ad-AKAP12A was subjected to micro-array analysis using Affymetrix chips. Interestingly, it was found that the expressions of several genes involved in cholesterol biosynthesis, including HMG-CoA reductase (HMGCR) and mevalonate pyrophosphate decarboxylase (MVD) were up- regulated more than 1.7-fold by AKAP12A overexpression (data not shown). To confirm this finding, we first exam- ined the mRNA levels of MVD and HMGCR at 24 h after Ad-AKAP12A infection. Fig. 1A shows that the mRNA levels of MVD and HMGCR were up-regulated by Ad-AKAP12A according to RT-PCR and real-time RT-PCR findings. We next searched for those transcription factors that controlled these genes. Expressions of these MVD and HMGCR have previously been shown to be regulated by SREBP-2 (Horton et al., 2002). Thus, we examined the effect of AKAP12A on SREBP activation, namely, on the cleavages of SREBP- 2 and SREBP-1. SREBP-2 was detected using an antibody recognizing its C-terminus, whereas SREBP-1 was detected using an antibody recognizing its N-terminus. As shown
in Fig. 1B, AKAP12A overexpression caused the activation of SREBP-2, as demonstrated by the presence of its cleaved product, whereas the activation of SREBP-1 was not remark- ably changed. These results show that AKAP12A controls the activation of SREBP-2 but not that of SREBP-1 in AGS cells.

3.2. AKAP12A overexpression induces SREBP-2 activation independent of apoptosis in hepatic cells

Next, we studied the molecular mechanism of SREBP-2 activation using hepatocellular carcinomas (HCCs), because the liver is capable of removing cholesterol from the blood circulation. The basal expressions of AKAP12A, AKAP12B, SREBP-1, SREBP-2, SCAP and INSIG-1 were examined in eight HCC cell lines (Supplementary Fig. 1). However, no close correlation was found between AKAP12A expression and basal SREBP-2 activation, implying that AKAP12A is not essential for basal SREBP-2 activation. The HCC cell line SNU-449 was chosen to investigate the mechanism of SREBP-2 activation in response to AKAP12A overexpression because these cells basally express AKAP12A and have good adenoviral vector transduction efficiencies. SNU-449 cells were exposed to Ad-vectors at increasing MOIs and times, and the results presented in Fig. 2A show that increas- ing AKAP12A expression significantly increased SREBP-2 activation in a dose-dependent manner. Adenoviral vector itself (Ad-empty) influenced the activation of SREBP-1 in SNU-449 cells. Fig. 2B shows that the amount of cleaved SREBP-2 by Ad-AKAP12A overexpression was significantly increased after 12 h, while the expression of SREBP-2 pre- cursor protein was augmented after 18 h. The expressions of the mRNAs encoding SREBP-2 target genes (i.e., HMGCR, LDLR and SREBP-2) were elevated in response to AKAP12A overexpression for 16 h. INSIG-1, which retains SCAP/SREBP complex in the ER (Yang et al., 2002), was rapidly induced by AKAP12A overexpression, as would be expected if SREBP was activated (Goldstein et al., 2006; Janowski, 2002). In contrast, SCAP expression remained unchanged, and AKAP12A did not modify the mRNA levels of S1P or S2P (data not shown). These results indicate that AKAP12A increases the activation of SREBP-2 protein, and that this is followed by upregulations in the transcriptions of SREBP- 2 target genes such as HMGGCR, LDLR, MVD, INSIG-1 and SREBP-2 itself.

If AKAP12A is important for SREBP-2 activation, endogenous AKAP12A depletion would be expected to inhibit SREBP-2 activation. Depletion of AKAP12A in SNU-449 cells was achieved by siRNA transfection, and SNU-449 cells depleted of AKAP12A showed SREBP-2 inactivation (Supplementary Fig. 2). These results confirm that the acti- vation of SREBP-2 is, at least in part, regulated by AKAP12A levels in SNU-449 cells.

According to a previous report (Higgins and Ioannou, 2001), SREBP-2 is cleaved by caspase-3 during apoptosis, and we and others reported that AKAP12 overexpression induces apoptotic cell death in gastric cancer and fibrosar- coma cells (Choi et al., 2004; Yoon et al., 2007). Thus, to exclude the possibility that SREBP-2 activation by AKAP12A is created by apoptosis, AGS and SNU-449 cells were treated with the caspase inhibitor z-VAD-fmk. In both cell lines, SREBP-2 activation was unchanged by z-VAD-fmk treatment (Supplementary Fig. 3), indicating that the regulation of SREBP-2 by AKAP12A is apoptosis-independent.

3.3. AKAP12A overexpression activates SREBP-2 in a SCAP-dependent manner

SCAP is important for the activation of SREBPs because it escorts them from the ER to the Golgi apparatus (Nohturfft et al., 2000). To investigate the involvement of SCAP in SREBP-2 activation by AKAP12A, we utilized 25-hydroxycholesterol (25-HC) and cholesterol, which are potent inhibitors of SCAP (Brown et al., 2002). Treatment of cells with sterols slightly increased the expression of AKAP12A (Fig. 3A). The addition of 25-HC and cholesterol to growth medium led to a marked reduction in SREBP-2 activation by AKAP12A. Moreover, when cells were treated with 25-HC or cholesterol, SREBP-2 activation by AKAP12 overexpression was also blocked (data not shown). Next, we addressed the same question using knock-down of SCAP expression by RNA interference (RNAi). Transfection of unspecific control siRNA oligonucleotides did not alter SREBP-2 activation by AKAP12A (Fig. 3B). However, this activation was suppressed by blocking SCAP with siRNA. We obtained similar results using different siRNA sets for SCAP (data not shown). Thus, these results indicate that SREBP-2 activation by AKAP12A requires SCAP.

3.4. Cholesterol-related changes in AKAP12A localization

It is believed that AKAP12 protein is present through- out cells, but that it shows a predilection for the plasma membrane (Lin et al., 2000b; Streb et al., 2004; Tao et al., 2006; Xia et al., 2001). Here, we investigated whether cholesterol regulates the localization of AKAP12A in cells. The hydrophobic amine U18666A blocks the release of cholesterol from late endosome to the membrane and ER, resulting in the lysosomal accumulation of cholesterol (Liscum and Faust, 1989). In addition, it is known that cholesterol is rapidly removed from the plasma membrane in response to methyl-beta cyclodextrin (MCD) treatment, and thus MCD has been extensively used as a choles- terol binding and depleting reagent (Marwali et al., 2003; Rodal et al., 1999). We first examined the expressions and localizations of endogenous AKAP12 in SNU-449 cells treated with U18666A; however, AKAP12 expression was found to be unchanged (Supplementary Fig. 4). Inter- estingly, U18666A-treated cells induced a dramatic shift of AKAP12 from the plasma membrane to the cytosol (Fig. 4A). Moreover, when SNU-449 cells were treated with MCD for 1 h, membrane-associated AKAP12 disappeared (Fig. 4B), whereas the expression of AKAP12 was unchanged (Supplementary Fig. 4B). To further investigate this, COS-7 cells were transiently transfected with GFP-Akap12α, incubated with U18666A for 24 h or MCD for 1 h, and processed for confocal microscopy using lysotracker, a lyso- somal marker (Fig. 4C). GFP-Akap12α was localized to the plasma membrane during the unstimulated condition, but after cells have been treated with U18666A, GFP-Akap12α moved from plasma membranes to the cytosol and lyso- somal membranes. On the other hand, when membrane cholesterol was depleted by adding MCD, GFP-Akap12α was re-localized from plasma membranes to the cytoplasm, but not to lysosomal membranes. We further confirmed AKAP12 localization using an antibody that recognizes human lysosome-associated membrane protein 2 (LAMP2). AKAP12 was found to co-localize with LAMP2 in U18666A- treated cells, but not in MCD-treated cells (Supplementary Fig. 5). These findings suggest that AKAP12 localizes to plasma membranes, cytoplasm, and lysosomal membranes in a cholesterol-dependent manner.

3.5. The N-terminal region of AKAP12A binds directly to cholesterol in vitro

A recent report suggested that the N-terminal myristoy- lation site in AKAP12A is not critical for AKAP12 localization to the plasma membrane (Tao et al., 2006). Rather, three small, positively charged basic domains (PCDs) encom- passing the basic/hydrophobic clusters of AKAP12A are important (Streb et al., 2004). It has been known that sterols can bind within the hydrophobic tunnels of several pro- teins including NPC2 and Osh4 (Im et al., 2005; Ko et al., 2003). In addition, based on the similar distributions of AKAP12 and cholesterol (Fig. 4), we checked whether AKAP12A is a cholesterol binding protein. Cholesterol bind- ing ability of AKAP12A was studied using His-AKAP12A proteins in vitro. Since 6 × His control peptide is very small, it is lost during purification. Thus, we selected Hpc2 as a non-specific control protein because this poly- comb protein functions as a SUMO E3 ligase irrespective of cholesterol (Kagey et al., 2003). To examine the poten- tial role of AKAP12A myristoylation in cholesterol binding, we first employed His-fusion proteins encoding the dele- tion mutant (17–1782) or the G2A mutant of the AKAP12A myristoylation site. As shown in Fig. 5A, His-wild type (1–1782) was found to bind to [3H]-cholesterol better than His-Hpc2. However, two mutants also bound to cholesterol, despite loss of the N-myristoylation motif. We next prepared His-fusion proteins encoding various deletion fragments of AKAP12A to identify the choles- terol binding domain of AKAP12A, and to map this domain (Supplementary Fig. 6). As reported by others (Tao et al., 2003), all of the purified proteins behave as dimeric. Using cholesterol binding assays, it was found that cholesterol interacts with the N-terminal region of AKAP12A but not with the C-terminal or the AKAP domain (Fig. 5B). The 1–165 fragment protein, which is deficient in all PCDs, was markedly poorer at binding cholesterol than the 1–181 or 1–294 fragments. To determine whether PCDs directly bind to cholesterol, we synthesized 6 × His-tagged PCD1, PCD2, and PCD3 peptides and measured their cholesterol binding affinities. As shown in Fig. 5C, all three peptides (especially PCD1) were found to have a higher binding affinity than the control peptide. Together, these results suggest that AKAP12A directly binds to cholesterol in vitro.

3.6. Effect of inducible AKAP12A overexpression on cholesterol efflux

Cellular cholesterol efflux is important for choles- terol homeostasis. To examine whether AKAP12A increases cholesterol efflux, SNU-449 cells were labeled with [3H]- cholesterol for 48 h in culture medium containing 10% FBS, and efflux of radioactive cholesterol to ApoAI was measured. During the 1 h efflux period of the [3H]- cholesterol to BSA plus ApoAI, [3H]-cholesterol efflux was significantly higher for AKAP12A-overexpressing cells than for PBS or control virus-treated cells (Fig. 6A). As for BSA, ApoAI, HDL, and 10% serum acceptors, AKAP12A overexpressing cells also showed an increase in choles- terol delivery to acceptors as compared with control virus-infected cells (Fig. 6B). These data suggest that AKAP12A enhances cellular cholesterol efflux from hepatic cells.

4. Discussion

In the present study, we characterize the effects of inducible AKAP12A overexpression on cellular cholesterol metabolism in hepatic cells. The work was initiated to iden- tify pathways regulated by AKAP12A. For this purpose, we analyzed changes in gene expression in response to AKAP12A overexpression using micro-arrays. Only a small number of genes were found to be regulated by AKAP12A in AGS gastric cancer cells, and even these showed only moderate expressional changes. Thus, micro-array results were insufficient to define the downstream signaling of AKAP12A. However, because several of the genes identi- fied are known to be related to cholesterol biosynthesis, we considered the micro-array results to be meaningful. Thus, we followed the micro-array analysis with RT-PCR and real- time RT-PCR. Fortunately, these analyses produced positive results (Fig. 1A).

In terms of the molecular mechanism whereby AKAP12A activates SREBP-2, it has been reported that sev- eral growth factors activate SREBPs via two major signaling mediators, MAPK and Akt (Demoulin et al., 2004; Du et al., 2006; Porstmann et al., 2005; Zhou et al., 2004). We checked the involvements of both, but AKAP12A still acti- vated SREBP-2 when cells were treated with PD98059 (a MAPK inhibitor) or LY294002 (an Akt inhibitor) (unpub- lished data). Likewise, we checked whether PKA is involved in SREBP-2 activation, because AKAP family members are scaffold proteins that recruit PKA. However, AKAP12A- mediated SREBP-2 activation was not inhibited by KT5720 (a PKA inhibitor; unpublished data).

Interestingly, treatment with U18666A or MCD dramatically changed the cellular localization of AKAP12. Moreover, the cellular localizations of AKAP12A and cholesterol were found to be similar, as follows. First, although SREBP-2 activation was rapidly induced by U18666A treatment (Supplementary Fig. 4A), the lyso- somal accumulations of AKAP12 and cholesterol were observed under the microscope to occur simultaneously after 12 h of U18666A treatment (unpublished data). Second, U18666A, which causes the lysosomal accumu- lation of cholesterol (Liscum and Faust, 1989), led to the lysosomal targeting of GFP-Akap12α. In the case of MCD, since it simply extracts cholesterol from the plasma mem- brane and disrupts caveolae, it induced a diffuse AKAP12A cytoplasmic distribution. These findings suggested that AKAP12A acts as a cholesterol-dependent scaffold protein.

AKAP12A has a putative N-myristoylation site and three

PCDs (Streb et al., 2004; Tao et al., 2006). In particular, these three PCDs comprise its basic/hydrophobic clusters. PCDs might also anchor AKAP12A to the negatively charged inner leaflet of the plasma membrane via electrostatic binding. AKAP12A does not possess a specific sterol binding/transfer domain, such as, the sterol sensing domain (SSD) found SCAP and HMGCR (Chang et al., 2006; Goldstein et al., 2006). However, sterol binding proteins, such as, NPC2 and Osh4, have hydrophobic tunnels which interact with sterols (Im et al., 2005; Ko et al., 2003). Moreover, Parton et al. suggested that caveolin-1, a cholesterol binding pro- tein (Murata et al., 1995), might bind to cholesterol using conserved basic and bulky hydrophobic residues (Parton and Simons, 2007). Our results show that the three PCDs of AKAP12A bind directly to cholesterol in vitro. To fur- ther confirm this finding, an in vivo examination using photoactivatable cholesterol (Ohgami et al., 2004) will be necessary.

It has been established that the control of both cholesterol biosynthesis and transport contributes to the maintenance of cellular cholesterol homeostasis. Cellu- lar levels of cholesterol are constantly monitored via the activation of SREBP-2. Low levels of cholesterol in the ER are recognized by SCAP, which activates SREBP-2. Thus, it is possible that AKAP12A may increase SREBP-2 cleavage by increasing cholesterol efflux. In fact, sev- eral efflux-related proteins, such as, hepatic ABCA1 and ORP2, increase cholesterol efflux, which lead to compen- satory changes in expressions of SREBP-2 target genes, such as, HMGCR and LDLR (Basso et al., 2003; Hynynen et al., 2005). Caveolin-1, a tumor suppressor in breast cancer (Williams and Lisanti, 2005), also regulate choles- terol efflux (Arakawa et al., 2000; Fielding et al., 1999), and the in vivo expression of caveolin-1 in mouse liver cause increases in plasma HDL levels (Frank et al., 2001). These findings underscore the central role that plasma membranes play in the regulation of cholesterol efflux. Our results show that AKAP12A overexpression increases cholesterol efflux to several extracellular acceptors, which is consistent with the findings of a previous study on ORP2 (Hynynen et al., 2005). Additional studies are required to characterize how AKAP12A increase cholesterol efflux.

In conclusion, the present study demonstrates for the first time that AKAP12A participates in cholesterol metabolism. Our results suggest a function of AKAP12A in both SREBP-2 activation and cholesterol trafficking.