Methyl-β-cyclodextrin Suppresses Hyaluronan Synthesis by Down-regulation of Hyaluronan Synthase 2 through Inhibition of Akt*

Hyaluronan synthases (HAS1–3) are integral plasma membrane proteins that synthesize hyaluronan, a cell surface and extracellular matrix polysaccharide necessary for many biological processes. It has been shown that HAS is partly localized in cholesterol-rich lipid rafts of MCF-7 cells, and cholesterol depletion with methyl-β-cyclodextrin (MβCD) suppresses hyaluronan secretion in smooth muscle cells. However, the mechanism by which cholesterol depletion inhibits hyaluronan production has remained unknown. We found that cholesterol depletion from MCF-7 cells by MβCD inhibits synthesis but does not decrease the molecular mass of hyaluronan, suggesting no major influence on HAS stability in the membrane. The inhibition of hyaluronan synthesis was not due to the availability of HAS substrates UDP-GlcUA and UDP-GlcNAc. Instead, MβCD specifically down-regulated the expression of HAS2 but not HAS1 or HAS3. Screening of signaling proteins after MβCD treatment revealed that phosphorylation of Akt and its downstream target p70S6 kinase, both members of phosphoinositide 3-kinase-Akt pathway, were inhibited. Inhibitors of this pathway suppressed hyaluronan synthesis and HAS2 expression in MCF-7 cells, suggesting that the reduced hyaluronan synthesis by MβCD is due to down-regulation of HAS2, mediated by the phosphoinositide 3-kinase-Akt-mTOR-p70S6K pathway.

Cholesterol is an important membrane constituent of mammalian cells, plasma membrane in particular. It decreases the fluidity of the membrane and is a major component of the detergent-resistant membrane microdomains called lipid rafts (1). Increased serum cholesterol and cholesterol accumulation in atherosclerotic lesions have been long associated to cardiovascular disease (2) and atherosclerosis (3,4), respectively. Abnormal cholesterol accumulation has been also found in some malignancies, such as prostate tumors (5,6). Furthermore, recent studies suggest that inhibition of the rate-limiting enzyme in cholesterol synthesis (3-hydroxy-3-methyl-glutarylcoenzyme A reductase) by long-term statin therapy decreases the incidence of certain tumor types (7,8).
Hyaluronan, a large glycosaminoglycan mainly present in the extracellular matrix of vertebrates, is composed of repeating disaccharide units containing glucuronic acid (GlcUA) and N-acetylglucosamine (GlcNAc). It is produced at the plasma membrane by three hyaluronan synthases (HAS1-3). HASs use UDP-GlcUA and UDP-GlcNAc as substrates and extrude the forming hyaluronan chain to the extracellular space (18).
The microenvironment of HAS, especially the phospholipid composition of the plasma membrane, is important for HAS activity (28,(37)(38)(39)(40). Cardiolipin is the best activator of hyaluronan synthesis in Streptococcus equi FM100 and exogenous cardiolipin was reported to restore the activity of HAS sup-pressed by the hyaluronan synthesis inhibitor 4-methylumbelliferone (28). Mammalian cells do not contain cardiolipin in plasma membrane (41), but depletion of cellular cholesterol by M␤CD suppresses hyaluronan synthesis in aortic smooth muscle cells (42) and the suppression can be reversed by re-addition of cholesterol (42). Moreover, aortic smooth muscle cells from hyperlipidemic rabbits and human skin fibroblasts from hypercholesterolemic patients secrete 2-to 4-fold more hyaluronan to culture medium than normolipidemic controls (42).
The aim of this work was to study the effect of M␤CD on hyaluronan synthesis and characterize its mechanism of action in MCF-7 human breast cancer cells. We found that M␤CD inhibits hyaluronan synthesis but does not change the molecular mass of newly synthesized hyaluronan, whereas it specifically suppressed the mRNA level of HAS2. This was associated with reduced phosphorylation of Akt and p70S6 kinase, suggesting that inhibition of the phosphoinositide 3-kinase (PI3K)-Akt pathway accounted for the down-regulation of HAS2, and was a major contributor to the suppression of hyaluronan synthesis exerted by M␤CD.
Cholesterol Quantification-Subconfluent cells were treated with different concentrations of M␤CD (0 -2.5 mM, Sigma) for 4 h, lysed with chloroform:methanol (2:1), and centrifuged for 5 min at 16,000 ϫ g. The organic phase was vacuum-dried overnight and the lipids were re-dissolved in 2-propanol containing 10% Triton X-100. Cholesterol quantification with a fluorometric assay was performed with the Cholesterol/Cholesteryl Ester Quantitation kit according to the manufacturer's instructions (Calbiochem). Briefly, the samples and the cholesterol standards (0 -1 g/well) were mixed with a reaction mixture containing cholesterol reaction buffer, cholesterol probe, and enzyme mixture and applied to 96-well plates. After a 1-h incubation at 37°C, avoiding light exposure, fluorescence was measured at an excitation wavelength of 530 nm and emission at 590 nm.
Hyaluronan Assay-MCF-7 cells (80,000) were seeded on 24-well plates, grown for 24 h, and changed into fresh medium (with 0.5-1% FBS) containing 0 -2.5 mM M␤CD for 6 and 24 h. Recovery of hyaluronan synthesis after M␤CD treatment was analyzed by quantifying hyaluronan accumulation to culture medium 0 -24, 24 -48, and 48 -72 h after removal of M␤CD. Lovastatin (10 -50 M, Sigma) and ␣-cyclodextrin (0 -2.5 mM, Sigma) were also incubated with MCF-7 cells for 24 h. For some cultures, 32 M cholesterol:M␤CD was added after a 24-h M␤CD treatment to refeed the cells with cholesterol. After a 2-h incubation, fresh growth medium containing 0.5% FBS was changed and the cells were allowed to synthesize hyaluronan for 6 h before collecting the medium and counting the cells. After 6-and 24-h incubations, the cells were counted and medium was analyzed for hyaluronan content using a sandwich-type enzyme-linked immunosorbent -like assay as described previously (43).
Molecular Mass of Hyaluronan-200,000 cells were seeded on 6-well plates and cultured for 48 h. Fresh medium containing 0.5-1% FBS, 95 Ci/ml of [ 3 H]glucosamine (PerkinElmer Life Sciences), and 0 or 2 mM M␤CD was added to the cells and incubated for 24 h. The culture medium was collected and stored at Ϫ20°C for analysis. The cells were trypsinized and centrifuged at 800 ϫ g for 5 min. The trypsin supernatant (pericellular hyaluronan) was stored at Ϫ20°C until analysis. The cell pellet was suspended in 100 mM ammonium acetate, pH 7.0, followed by addition of proteinase K (Sigma) (250 g/sample) and incubation for 2 h at 60°C to degrade proteins. The proteinase was inactivated by boiling the samples for 10 min and the samples were centrifuged at 16,000 ϫ g for 5 min at 4°C. The supernatants, containing the intracellular hyaluronan, were stored at Ϫ20°C until analysis.
The increase of [ 3 H]glucosamine in the supernatant of the hyaluronidase-treated aliquot was a specific measure of hyaluronan. The void volume of the Sephacryl S-1000 column (V 0 ) was considered to be in the earliest hyaluronan-positive fraction for a standard hyaluronan (Hyalose 2,500 kDa, Hyalose, L.C.C., Oklahoma City, OK). The total volume (V t ) was considered to be at the elution position of glucuronic acid. The size distribution of hyaluronan in the samples was estimated from the K av values of the peak fractions of known hyaluronan standards (150, 500, and 2,500 kDa), provided by Hyalose.
Extraction of UDP-nucleotide Sugars-Cells (800,000) were seeded on 8.6-cm plates and cultured for 24 h, then incubated with 0, 0.25, 0.5, 0.75, 1, and 2.5 mM M␤CD in medium containing 1% FBS for 24 h. For anion-exchange HPLC the cells were washed with cold phosphate-buffered saline on ice and 80 l of 0.1 mM methyl adenosine 5Ј-triphosphate (Sigma) was added for an internal standard. Next, 3 ml of cold acetonitrile was added to precipitate proteins and extract the UDP-sugars, followed by the addition of 1 ml of deionized water. The cells were scraped off and after centrifugation the supernatant was transferred to a clean tube, evaporated in a vacuum centrifuge, and dissolved by vortexing into 400 l of Milli-Q water for anion exchange HPLC.
Quantitation of UDP-GlcUA and UDP-GlcNAc by Anion Exchange HPLC-A CarpoPac TM PA1 column, 4 ϫ 250 mm (Dionex, Sunnyvale, CA), was eluted at 1 ml/min. An aliquot of the extract (100 -200 l) was fractionated with a gradient made of 1 mM NaOH (A) and 1 M sodium acetate in 1 mM NaOH (B) using detection at 260 nm. The column was equilibrated with a 80:20 (v/v) mixture of A and B. Elution was performed with the following percentages of B: T 0 ϭ 20%; T 10 ϭ 55%; T 25 ϭ 55%; T 35 ϭ 80%; T 40 ϭ 100%; T 50 ϭ 100%. For quantitation the integrated peak areas of the UDP-sugars were compared with that of the internal standard.
Quantitative RT-PCR-MCF-7 cells (100,000) were seeded on 6-well plates and grown for 24 h. The cells were treated with 0 -2.5 mM concentrations of M␤CD in medium containing 0.5-1% FBS for 2, 4, and 24 h after which total RNA was extracted with EUROzol (Euroclone) and the remaining DNA was removed with TURBO DNA-free TM (Ambion, Austin, TX). Equal amounts of RNA were reverse transcripted and real time PCR was performed with a MX3000P thermal cycler (Stratagene, La Jolla, CA) using Brilliant SYBR Green Q-PCR master mix (Stratagene) and specific primers for HAS2, HAS3, CD44, HYAL2, and acidic riboprotein P0 (ARP0) (44) ( Table 1). Fold-inductions were calculated using the formula 2 Ϫ⌬⌬Ct , where ⌬⌬C t stands for the ⌬C t (treatment) Ϫ ⌬C t (control). ⌬C t is C t of HAS2, HAS3, CD44, HYAL2 Ϫ C t of ARP0, and C t is the cycle where the detection threshold is crossed.
Phosphokinase Array-A phosphokinase array (R&D Systems, Abingdon, UK) was used to analyze the phosphorylation status of 46 kinases and transcription factors. 1.3 million cells were plated in 8.6-cm dishes and grown for 24 h. Fresh medium containing 0.5% FBS was changed 24 h prior to treatments to ensure similar metabolic states in the cells. 1 mM M␤CD was added for 10 min, 30 min, 2 h, and 4 h. The cells were then solubilized at 1 ϫ 10 7 cells/ml in lysis buffer 6 provided in the kit and incubated on ice for 30 min, with intermittent gentle rocking, and centrifuged at 16,000 ϫ g for 5 min at 4°C, after which the supernatant was transferred to a clean tube. Protein concentration was determined by Bradford's assay and the samples were stored at Ϫ70°C until the analysis. 200 g of protein was used for each assay performed according to the manufacturer's instructions. The density of the spots in an exposed film was analyzed by Image J software. Background signal (negative control value) was subtracted from each captured spot and the average signal value was calculated as a mean of duplicate spots representing each phosphorylated protein.
Western Blotting-1.3 million cells in 8.6-cm dishes were grown until they reached 80% confluence. Fresh medium containing 0.5% serum and 0 and 1 mM M␤CD was changed for 2, 4, and 24 h after which cytosolic (pAkt and phospho-sterol-regulated element-binding protein-2 (SREBP-2)), membrane (pSREBP-2), and nuclear proteins (pSREBP-2) were extracted using the Qproteome kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. The plates were placed on ice, medium was removed, and the cells were rinsed with phosphate-buffered saline. The cells were scraped off to 1.5 ml of phosphate-buffered saline, transferred into 2.0-ml microcentrifuge tubes, and centrifuged at 500 ϫ g for 10 min at 4°C. The supernatant was discarded and the pellet was washed and centrifuged as above. The pellet was suspended in ice-cold extraction buffer (CE1) provided in the kit, followed by incubation for 10 min at 4°C with intermittent mixing. After centrifugation at 1,000 ϫ g for 10 min at 4°C the supernatant containing the cytosolic proteins were frozen at Ϫ70°C. The pellet was resuspended in extraction buffer CE2, followed by a 30-min incubation at 4°C and centrifugation at 6,000 ϫ g for 10 min at 4°C. Supernatant containing the membrane proteins was stored at Ϫ70°C. To extract the nuclear proteins the pellet was resuspended in Benzonase Nuclease and incubated for 15 min at room temperature. After addition of extraction buffer CE3 and incubation at 4°C for 10 min, the samples were centrifuged at 6,800 ϫ g for 10 min at 4°C. Nuclear proteins found in the supernatant were frozen at Ϫ70°C until analysis.
Equal amounts of the samples were mixed with 4ϫ SDS buffer, separated on 10% SDS-PAGE, and transferred onto Immobilon TM -NC membranes (Millipore, Bedford, MA) by 135 mA/gel constant current with a Biometra blotter (Göttingen, Germany). The membranes were blocked in 10 mM Tris, 150 mM NaCl, pH 7.4 (TBS), containing 5% fat-free milk powder for 30 min at room temperature, then incubated with a pAkt antibody (Cell Signaling, Danvers, MA) diluted 1:250 in 1ϫ TBS containing 5% fat-free milk powder overnight at 4°C.
The membranes were scanned for fluorescence with Odyssey reader (LI-COR, Lincoln, NE). After scanning, the pAkt antibodies were removed by 0.2 M NaOH treatment for 5 min at room temperature followed by incubation with an antibody against total Akt (diluted 1:500, Cell Signaling) in TBS containing 5% fat-free milk powder overnight at 4°C. After washes, the blots were incubated with goat anti-rabbit IgG, DyLight TM 680 secondary antibody 680 (1:500 dilution in TBS containing 5% fat-free milk powder, ThermoScientific) for 1 h at room temperature. The pAkt level in each sample was normalized to the total Akt in the sample.
Caspase Staining-MCF-7 cells seeded on 8-well chamber slides (Nalge Nunc, Naperville, IL) precoated with type I collagen (BD Biosciences) were treated for 4 and 24 h with 0 and 1 mM M␤CD. After treatments, slides were fixed with 2% paraformaldehyde, permeabilized with 0.1% Triton X-100 in 1% bovine serum albumin, and stained overnight at 4°C with anti-caspase 3 antibody (Cell Signaling, Danvers, MA) diluted 1:150 in 1% bovine serum albumin. Primary antibody was detected with biotinylated antirabbit secondary antibody (Vector Laboratories) diluted 1:200 in phosphate buffer. This was followed by a 1-h incubation in avidin-biotin-peroxidase (ABC-standard kit, Vector Laboratories), and the color was developed with 0.05% 3.3Ј-diaminobenzine and 0.03% H 2 O 2 for 5 min. The nuclei were stained with Mayer's hematoxylin.
Statistical Analysis-The significance of differences between the groups in hyaluronan assays, analysis of UDP-sugars, and quantitative HAS RT-PCR experiments were tested using univariate analysis of variance and Tukey's post hoc test. Statistical significance was set p Ͻ 0.05. Statistical analysis was performed when the experiment was repeated three times or more.

M␤CD Depletes Cholesterol and Suppresses Hyaluronan
Secretion in MCF-7 Cells-Several concentrations of M␤CD were applied to explore its cholesterol-depletive effect in this cell line. M␤CD at 0.5, 0.75, 1.0, and 2.5 mM concentrations decreased cell-associated cholesterol in 4 h by 34.7, 44.3, 57.6, and 64.4%, respectively (Fig. 1A). MCF-7 cells secreted into growth medium ϳ0.7 ng of hyaluronan/10,000 cells in 6 h, and ϳ1.5 ng in 24 h. Treatment for 6 h with 0 -2.5 mM M␤CD in medium containing 1% FBS did not show significant changes in hyaluronan secretion (data not shown), whereas when incubation was extended to 24 h hyaluronan in the growth medium was decreased by ϳ36% (Fig.  1B). However, when cells were first treated with M␤CD for 24 h, then changed back to medium without M␤CD, hyaluronan secretion during a 6-h follow-up was still decreased by 81% compared with untreated cells (Fig. 1C).
Recovery of the cells was studied by analyzing hyaluronan secretion 0 -72 h after M␤CD (0, 1, and 2.5 mM concentrations) had been removed from the medium (Fig. 2). Interestingly, after a 24-h culture in the medium containing 0.5% FBS hyaluronan production was still suppressed by 68 -77% ( Fig. 2A), whereas 24 -48 h after removal of M␤CD, the cells were partly recovered, producing hyaluronan 75% of the level of controls (Fig. 2B). After 48 -72 h the MCF-7 cells were fully recovered, and culture medium collected from M␤CD-treated cells contained hyaluronan even more (161-257%) than untreated cells (Fig.  2C). A possible increase of cell death in response to M␤CD was analyzed using immunostaining against caspase 3. The results showed that there was no increase in the amount of apoptotic cells (data not shown). Our results suggests that cholesterol depletion has a long-term suppressive effect on hyaluronan synthesis. However, the cells remain viable although their hyaluronan synthesis requires a relatively long time to fully recover.
To confirm that cholesterol depletion rather than the chemical nature of M␤CD causes suppression of hyaluronan synthesis, the cells were incubated for 24 h with ␣-cyclodextrin (0 -2.5 mM) that does not affect cholesterol content. It had no effect on hyaluronan synthesis (data not shown), suggesting that the effects of M␤CD were due to its influence on cellular cholesterol. The cells were also treated with 1-10 M lovastatin to suppress endogenous cholesterol synthesis. This had no effect on hyaluronan synthesis in a 24-h incubation (data not shown).
To rescue the synthesis of hyaluronan, 32 M cholesterol loaded onto M␤CD was added for 2 h after a 24-h treatment with 1 mM M␤CD, followed by 6 h incubation in fresh growth medium (0.5% FBS). During the 6-h follow-up cholesterolloaded M␤CD increased hyaluronan secretion by 55% as compared with cholesterol-depleted cells (Fig. 1C). Higher cholesterol:M␤CD concentrations could not be used because of insolubility.
M␤CD Does Not Change the Size of Hyaluronan Produced by MCF-7 Cells-To reveal whether cholesterol depletion of the plasma membrane disturbs HAS stability, as suggested to influence the amount and size of newly synthesized hyaluronan (45), we studied the molecular mass distribution of hyaluronan in intracellular, pericellular, and medium compartments after 24 h metabolic labeling with and without 1 mM M␤CD (Fig. 3). M␤CD decreased 3 H-labeled hyaluronan in culture medium by 74% (Fig. 3A), whereas on the cell surface (Fig. 3B) and within the cells (Fig. 3C) the medium was suppressed by 78 and 93%, respectively. The mean molecular mass of secreted and pericellular hyaluronan was ϳ1,500 kDa (Fig. 3, A and B), whereas hyaluronan inside the cell was smaller and polydisperse, with a mean molecular mass of 500 kDa. Despite decreased synthesis by 2 mM M␤CD, the relative distribution of the molecular mass was not significantly changed in any of the culture compartments in MCF-7 cells (Fig. 3). Therefore, cholesterol depletion in MCF-7 cells did not disturb HAS activity in a way that would lead to production of shorter hyaluronan molecules.
Effect of M␤CD on Cellular UDP-GlcUA and UDP-GlcNAc-It has been recently shown that the cytosolic contents of HAS substrates UDP-GlcUA and UDP-GlcNAc can limit hyaluronan synthesis (36,44). The contents of these hyaluronan precursors were therefore analyzed as one potential alternative causing the inhibition of hyaluronan synthesis exerted by M␤CD. HPLC by anion-exchange chromatography showed that untreated MCF-7 cells contained on the average 21 pmol of UDP-GlcUA and 269 pmol of UDP-GlcNAc/10,000 cells 24 h after a change of fresh growth medium (Fig. 4). Instead of decreasing the levels UDP-GlcUA and UDP-GlcNAc, 1 and 2.5 mM M␤CD had a tendency to slightly increase the concentration of both HAS substrates (Fig. 4, A and B). The data indicate that the suppression of hyaluronan synthesis cannot be due to depletion of UDP-sugar precursors. The enlarged pools of UDP-GlcNAc lead to greater dilution of the [ 3 H]glucosamine used in radiolabeling, reducing the specific activity of UDP-  GlcNAc, and explaining the apparently stronger depletion of hyaluronan in Fig. 3, than expected on the basis of direct quantitation by the enzyme-linked sorbent assay in Fig. 1B.
M␤CD Down-regulates HAS2 mRNA Level-The effect of 0 -2.5 mM M␤CD on HAS, HYAL, and CD44 genes was studied by quantitative RT-PCR after 2-, 4-, and 24-h treatments with M␤CD. The main HAS expressed in MCF-7 cells was HAS2. HAS1 was not expressed (data not shown), and HAS3 was present at a low level. M␤CD decreased HAS2 mRNA at all time points (Fig. 5, A-C). After a 24-h treatment with 0.5, 1.0, and 2.5 mM M␤CD there was a dose-dependent, 36 -89% down-regulation of HAS2 expression (Fig. 5C). In contrast, there was no consistent change in the mRNA level of HAS3 (Fig. 5, D-F).
On the catabolic side, the MCF-7 cells did not express significant quantities of HYAL1 (data not shown), and the mRNA levels of HYAL2 and the hyaluronan receptor CD44 were not affected by cholesterol depletion (data not shown), suggesting that the effect of M␤CD on HAS2 expression was rather specific, and its influence on hyaluronan content in the medium was due to a decreased rate of synthesis.
Effect of Signaling Inhibitors on Hyaluronan Synthesis-To find out how HAS2 is regulated in MCF-7 cells, and to learn about the signals M␤CD might use in HAS2 repression, we screened the effect of several known signaling inhibitors on hyaluronan production. During a 24-h incubation period, the PI3K inhibitors wortmannin (1 M) and LY294002 (25 M) suppressed hyaluronan secretion by 47 and 62%, respectively, whereas an inhibitor of Akt (1 M), downstream of PI3K, decreased hyaluronan secretion by 65% (Fig. 6A). Inhibition of mTOR, a downstream target of Akt, suppressed hyaluronan production by 46%, suggesting that the PI3K-Akt-mTOR pathway is involved in the regulation of hyaluronan synthesis (Fig. 6A).
To sum up, the results in MCF-7 cells support earlier findings published by many research groups that several signaling pathways (23, 46 -48), but especially the PI3K-Akt cascade (23,49) are involved in regulation of hyaluronan synthesis. Inhibition of PI3K and Akt Down-regulates HAS2 mRNA Levels-The PI3K-Akt pathway inhibitors, being most effective in blocking hyaluronan synthesis, were checked for their influence on the expression of HAS2, the main enzyme responsible for hyaluronan synthesis in these cells. Quantitative RT-PCR experiments revealed that 4-h treatments with wortmannin and LY294002 suppressed HAS2 mRNA levels by 61 and 79%, respectively, whereas the Akt VIII inhibitor down-regulated HAS2 by 65% (Fig. 6B). The results suggest that the PI3K-Aktmediated signals are involved in the reduction of hyaluronan synthesis through down-regulation of HAS2 in MCF-7 cells.
Effect of M␤CD on Akt and p70S6K Phosphorylation-Cholesterol depletion by M␤CD has been suggested to cause changes in the phosphorylation of signaling molecules like ERK and Src (11,15). Using a phosphokinase array we screened the phosphorylation profiles of 46 kinases and their substrates to find out if M␤CD affected the signaling pathways important in HAS2 regulation. 1 mM M␤CD was added for 10 min, 30 min, 2 h, and 4 h.
The array revealed that 1 mM M␤CD increased phosphorylation of Akt at Ser-473 by 34% after a 10-min treatment, then returned to the control level (Fig. 7A). Phosphorylation of Akt at Thr-308 was also increased, but started to decrease after 30 min and was 65% lower than in controls by 4 h (Fig. 7A). Phosphorylation of the p70S6 kinase (p70S6K), located downstream of Akt and mTOR, was also affected by M␤CD. Although phosphorylation of p70S6K at Thr-229 was not changed consistently over time, increased phosphorylation was first noted at Thr-421 and Ser-424, with a peak at 30 min, followed by a decrease toward 4 h (Fig. 7B). The pattern was similar to the time-dependent increase and decline of Akt Thr-308. The decrease of pAkt at 4 h was confirmed with Western blotting (Fig.  7C). These results are in line with the idea that cholesterol depletion suppressed the Akt-p70S6K pathway after a 2-h treatment, and this down-regulated HAS2 expression and hyaluronan synthesis at later time points.
Because transcription factor SREBP-2 has two potential binding sites in the HAS2 promoter and has been suggested to mediate the effect of M␤CD (42), its possible role in the system was studied using immunofluorescence stainings and Western blots. However, M␤CD did not change the nuclear localization of SREBP-2 in MCF7 cells whether examined by staining of the cultures, or Western blots of the isolated cellular subcompartments including the nucleus (data not shown).

DISCUSSION
In the present study, we demonstrated that depletion of cellular cholesterol by the water-soluble cyclic oligosaccharide M␤CD reduces hyaluronan synthesis in MCF-7 breast cancer cells. M␤CD extracts cellular cholesterol from the plasma membrane into its hydrophobic cavity (50), thus changing membrane lipid composition and resulting, for example, in lipid raft disruption (10). Despite the profound change in membrane properties, no evidence was obtained for direct destabilization of HAS function in the plasma membrane, with molecular mass of the newly synthesized hyaluronan as an indicator. Instead, incubation with M␤CD down-regulated HAS2 expression without affecting CD44 or HYAL2. M␤CD reduced phosphorylation of Akt at Thr-308, and p70S6K at Thr-389 and Thr-421/Ser-424, both Akt and p70S6K being members of the PI3K-Akt-mTOR-p70S6K pathway. Because other inhibitors acting on this pathway suppressed HAS2 expression and reduced hyaluronan synthesis in MCF-7 cells, we conclude that the influence of M␤CD was due to this signaling downstream of PI3K.
Cholesterol and Hyaluronan Synthesis-Cholesterol is an important component of the plasma membrane, enriched in the heterogenous and dynamic lipid raft microdomains that have been suggested to regulate the function of multiple plasma membrane proteins (51,52). We have previously shown that overexpressed GFP-HAS3 is partly colocalized in lipid rafts and that M␤CD removes the hyaluronan synthesis-induced microvilli, a site for active hyaluronan synthesis.
In line with the present findings M␤CD inhibits hyaluronan synthesis in arterial smooth muscle cells (42). Aortic smooth muscle cells from low density lipoprotein receptor-deficient hyperlipidemic rabbits secrete more hyaluronan than cells from control rabbits (42), and diet-induced hypercholesterol-emia up-regulates HAS1 expression in experimental vein grafts (53). Hyaluronan is also accumulated in lipid-enriched areas of human atherosclerotic lesions (54) and in animal models of vascular injury and atherosclerosis (54 -56). This, together with the fact that overexpression of HAS2 accelerates development of atherosclerosis in apolipoprotein E-deficient mice (57), suggests a connection between hyaluronan and cholesterol metabolism.
The recovery from the M␤CD-induced suppression on hyaluronan synthesis was generally slow in MCF-7 cells, even after cholesterol addback, whereas arterial smooth muscle cells restore their hyaluronan synthesis after a 2-h cholesterol addback (42). The relatively slow response we found both in the initiation of inhibition and the recovery of hyaluronan synthesis is in line with a transcriptional mechanism of action rather than more direct effects. The fact that inhibition of endogenous cholesterol synthesis by lovastatin, blocking hydroxymethylglutaryl-coenzyme A reductase, did not influence hyaluronan production in MCF-7 cells or in normolipidemic smooth muscle cells (42), suggests that a relatively strong impact of cellular cholesterol is needed to change hyaluronan synthesis.
The finding that cholesterol depletion by M␤CD did not change the molecular mass distribution of hyaluronan in MCF-7 cells supports the idea that M␤CD does not cause a major disturbance of the HAS microenvironment that would lead to premature chain termination. The results were somewhat unexpected because phospholipid composition has been reported to be important for HAS activity in bacterial cells (28,(37)(38)(39)(40), where cardiolipin (not present in the mammalian plasma membrane (41)) is the best activator of hyaluronan synthesis and restores HAS activity also after 4-methylumbelliferone treatment (28). However, by disrupting microvilli (43) M␤CD could reduce the plasma membrane surface area available for HAS, and therefore indirectly inhibit hyaluronan synthesis in those cells expressing particularly high levels of HAS.
Effect of M␤CD on Cellular UDP-sugars-Suppression of hyaluronan synthesis can be achieved by depleting the intracellular pool of UDP-GlcUA (44) or UDP-GlcNAc (36). The effect of M␤CD was, however, not mediated through this mechanism, as it actually slightly increased both precursors. Decreased membrane cholesterol has been reported to stimulate glucose transport (58), and increase GLUT-1 protein level (59) and GLUT-4 translocation to the plasma membrane (60) in various cell types, probably explaining the present finding. Transcriptional regulation rather that substrate supply thus appears to be the decisive factor in the effect of M␤CD on hyaluronan synthesis in MCF-7 cells.
Regulation of HAS2-Hyaluronan synthesis and HAS expression are subject to various regulatory signals in different cells (23,46,61). Among the inhibitors tested, those of PI3K (wortmannin and LY294002) and Akt (Akt inhibitor VIII) caused the strongest suppression of hyaluronan synthesis and HAS2 expression in MCF-7 cells. Inhibition of mTOR, a downstream effector of Akt, also lowered hyaluronan synthesis, supporting the involvement of the PI3K-Akt pathway in regulation of hyaluronan synthesis and HAS2, as schematically illustrated in Fig. 8.
Several previous reports agree with the present finding that the PI3K-Akt pathway influences HAS2 expression and hyaluronan synthesis, often through growth factors (62). Inhibition of PI3K by LY294002 suppresses platelet-derived growth factor receptor, ␤-polypeptide-induced HAS2 expression and hyaluronan production in dermal fibroblasts (23), and up-regulation of HAS2 and hyaluronan production was partly mediated by the activation of Akt in human osteoarthritic fibroblast-like synoviocytes (61).
M␤CD and Signaling-Extraction of cellular cholesterol by M␤CD influences the function of several signaling proteins. Epidermal growth factor receptor (11), human ErbB2 (11), ERK (11)(12)(13), p38 (11,14), Src (15), and ␤-catenin (16) are activated by M␤CD treatment in various cell types. Like other cell lines, MCF-7 cells showed increased ERK phosphorylation following M␤CD treatment. Inactivation of ERK inhibits, whereas activation of ERK stimulates hyaluronan synthesis, as shown by a number of studies (23,46,61). Therefore, our results suggest that activation of ERK does not mediate the suppression of HAS2 and hyaluronan synthesis in M␤CD-induced responses in MCF-7.
In MCF-7 cells, M␤CD treatment decreased the phosphorylation of Akt and p70S6K. This finding receives some support by papers showing that the M␤CD derivative 2,6-di-O-M␤CD suppresses PI3K-Akt signaling in macrophages (63), whereas the cholesterol oxidation product cholesterol-3-␤,5-␣,6-␤-triol activates the same pathway in endothelial cells (64). Furthermore, diet-induced hypercholesterolemia is associated with activation of Akt, mTOR complexes, and p70S6K in the heart of Yucatan pigs (65).
NF-B, a transcription factor and downstream target of Akt, has functional binding sites in the promoter of HAS2 and medi- ates up-regulation of HAS2 in keratinocytes in response to tumor necrosis factor ␣ (32). However, in MCF-7 cells inhibition of NF-B had no effect on the basal hyaluronan synthesis rate or HAS2 expression, suggesting that NF-B is unlikely to mediate the effect of M␤CD on HAS2 expression and hyaluronan synthesis.
In adipocytes, cholesterol depletion has been reported to increase the expression of SREBP-2 (59), which is regulated by the PI3K-Akt-mTOR pathway (66,67). However, our data indicated that the effect of M␤CD on HAS2 was not mediated by SREBP-2.
In the present study, we show that the abduction of cholesterol by M␤CD reduces hyaluronan synthesis in MCF-7 cells due to down-regulation of HAS2 that was mediated by the PI3K-Akt pathway (Fig. 8). This new information about the regulatory mechanisms of hyaluronan synthesis may aid in finding molecular targets to prevent excessive hyaluronan accumulation.