HDL3 Stimulates Multiple Signaling Pathways in Human Skin Fibroblasts
Abstract The influence of HDL3 on phospholipid breakdown was examined in human skin fibroblasts. HDL3 elicited phosphatidylcholine (PC) and phosphatidylinositol (PI) turnover and activated multiple phospholipases. In [14C]lyso-PC–labeled or [14C]choline (Cho)-labeled cells, a biphasic activation of PC-specific phospholipase D (PLD) with peak maxima 30 to 60 seconds and 5 to 7 minutes after stimulation with 20 μg/mL HDL3 was shown by (1) a 1.5- to 3-fold increase in [14C]phosphatidic acid (PA), (2) a 1.6-fold increase in Cho release, and (3) transphosphatidylation of PC to phosphatidylbutanol in the presence of 0.3% butanol. Activation of PC-specific PLD was paralleled by an activation of PC-specific phospholipase C (PLC). A significant increase in [14C]diacylglycerol (DG) was seen from 2 minutes after stimulation onward and remained for at least 2 hours. By means of butanol, the PA-phosphohydrolase (PPH) inhibitor propranolol, and the PC-PLC inhibitor D609, we demonstrated that the initial PC-derived DG formation occurred primarily by a coupled PLD/PPH pathway and that a major part of the sustained DG formation was derived directly from PC by PC-PLC. By downregulating protein kinase C (PKC) we demonstrated that PKC activates PC-PLC and desensitizes PC-PLD at longer incubation times. The sustained PC hydrolysis as well as HDL3-mediated PI turnover and PC resynthesis was observed on stimulation with 5 to 75 μg/mL HDL3, whereas the rapid activation of PC-PLD/PPH was detected only on stimulation with HDL3 at concentrations of between 10 and 75 μg/mL. Only the latter response could be mimicked by apolipoprotein A-I and apolipoprotein A-II proteoliposomes, and only this response was inducible by cholesterol loading. The HDL3-mediated second-messenger responses were inhibited by modification of HDL3 by tetranitromethane and could not be mimicked by protein-free liposomes. These data suggest that HDL3-induced cell signaling in human skin fibroblasts is mediated by specific protein-receptor interaction and that more than one agonist activity may be involved.
- Received May 5, 1995.
- Accepted June 15, 1995.
Epidemiological studies have identified low plasma HDL concentrations as a major coronary risk predictor (reviewed in Reference 11 ). It has been proposed that HDL is antiatherogenic by virtue of its potential to promote the efflux of cholesterol from the artery wall by two different mechanisms: (1) a nonspecific exchange of cholesterol molecules between HDL particles and the plasma membrane2 3 and (2) specific binding to high-affinity binding sites on the cell surface with mobilization of intracellular cholesterol.4 5 6 7 8 9 Another metabolic response that has been attributed to HDL is a strong mitogenic effect, which has been observed for various cell types in culture, such as vascular smooth muscle cells,10 11 vascular endothelial cells,12 fibroblasts,11 mammary epithelial cells,13 lymphocytes,14 and several cancer cell lines.15 16 17 18
A variety of agonists mediate intracellular metabolic responses by binding to specific cell-surface receptors, with subsequent breakdown of inositol phospholipids (reviewed in Reference 1919 ). Receptor-linked activation of PLC directed toward PIP2 leads to the liberation of IP3 and DG. IP3 causes intracellular calcium release and DG activates PKC, which in turn phosphorylates intracellular proteins. The second-messenger DG can also be produced by hydrolysis of PC (reviewed in Reference 2020 ). The hydrolysis of PC by PC-specific PLC produces ChoP and DG; the action of PC-specific PLD forms Cho and PA. PA can be converted to DG by PA-PPH but can itself also induce important second-messenger responses.
Previous data in the literature concerning the effects of HDL on the activation of cellular phospholipases have been inconsistent. The disparity may have arisen due to HDL heterogeneity, the use of different cell types, concentration-related effects, and experimental differences. HDL3-mediated PI turnover could be demonstrated in cultured human vascular smooth muscle cells.21 However, PI turnover was not observed in bovine aortic endothelial cells.22 In human platelets, low concentrations of HDL3 were reported to stimulate PC hydrolysis, indicating that PC-specific phospholipase C or D may be involved in HDL-mediated second-messenger responses.23 PC breakdown with no change in inositol lipids has been reported to occur in adipose cells.24 The strongest argument for specific HDL receptor-mediated processes has been presented by Oram and coworkers,4 5 7 who found that the stimulatory action of HDL3 on cholesterol mobilization in human skin fibroblasts is mediated by reversible binding of high-density apolipoproteins to high-affinity binding sites on the cell surface. In addition, we recently demonstrated that the HDL3-mediated mobilization of cellular cholesterol is impaired in fibroblasts from patients with familial HDL deficiency (Tangier disease), indicating that specific cholesterol efflux mechanisms may be important determinants of the plasma HDL cholesterol level.9 It was further demonstrated that the HDL3 receptor-mediated translocation efflux of intracellular cholesterol in fibroblasts occurs through activation of PKC, which increases twofold in the cell membrane 10 to 15 minutes after exposure of cells to HDL3.8 Moreover, the exposure of human fibroblasts to HDL3 is associated with a rapid and transient increase in cytosolic Ca2+.25 It has therefore been postulated that apolipoprotein binding transduces intracellular signals that in turn activate the cholesterol translocation pathway, probably through phosphorylation of key transport proteins.
To further characterize the underlying mechanism of HDL3-mediated cellular responses, we examined the influence of HDL3 on phospholipid breakdown in cholesterol-loaded human skin fibroblasts.
l-Lyso-3-phosphatidylcholine, [1-14C]palmitoyl (56 mCi/mmol), [methyl-14C]choline chloride (53 mCi/mmol), and [1-14C]AA were obtained from Amersham; and [32P]orthophosphoric acid (8500 to 9120 Ci/mmol), from NEN Du Pont. Autoradiography was performed by using Kodak X-OMAT film (Eastman Kodak). Media components were obtained from Flow. Plastic culture dishes were purchased from Falcon Labware. Scintillation-counting mixture (Ultima-Gold) was obtained from Canberra-Packard. Aluminum-backed silica gel 60 TLC plates and the solvents for TLC were purchased from Merck. The solvents for TLC were of reagent grade. All other reagents were obtained from Sigma Chemical Co and were of the highest purity available.
Human skin fibroblasts cultured from biopsies of adult human hip skin were grown and maintained in DMEM containing 10% FCS, 2 mmol/L l-glutamine, and 1% antibiotic/antimycotic solution (Sigma). Once separated, the dermis was cut into small pieces (0.5 mm on each side) and placed in a flask in DMEM. When these primary cultures were confluent they were expanded by passage. For experiments, cells between passage levels 3 and 6 were seeded in 60-mm culture dishes at a density of ≈75 000 cells per dish. At the state of 70% to 80% confluence, the cultures were cholesterol loaded to upregulate HDL3 binding.5 Cell layers were rinsed twice with PBS (pH 7.4) and incubated for 48 hours at 37°C in DMEM supplemented with free cholesterol (50 μg/mL) complexed with 10% FCS. To block PPH, cells were treated with 100 μmol/L propranolol for the final 3-hour incubation period in another experiment.26 To inhibit the PI-specific PLC, cells were treated with 15 μmol/L U 73122 (Calbiochem; dissolved and briefly sonicated in DMSO at 37°C) for 15 minutes and for the final 3-hour incubation period. To inhibit the PC-specific PLC, cells were treated with 30 μg/mL D609 (Sigma; 10 mg/mL stock solution dissolved in 80% acetone) for 30 minutes and for the final 3-hour incubation period. In one experiment, cells were incubated with 200 nmol/L PMA (Sigma; stock solution: 4 mg/mL dissolved in DMSO) or the inactive phorbol ester phorbol 12,13-dibutyrate (Sigma) at the final 24-hour period of cholesterol loading.27
HDL3 (d=1.125 to 1.210 g/mL) was isolated by standard differential ultracentrifugal flotation from fresh normal human plasma and dialyzed against 0.3 mmol/L Tris-HCl buffer (pH 6.8) containing 0.15 mol/L NaCl.28 The apoE constituted less than 0.3% of the total HDL3 protein. Modified HDL3 (HDL3-TNM) was prepared as described previously.29
Preparation of ApoA-I and ApoA-II Proteoliposomes
ApoA-I and apoA-II proteoliposomes containing normal apoA-I or apoA-II and DMPC were obtained by the cholate dialysis method according to Chen and Albers.30 ApoA-I was prepared from HDL3 by a procedure described previously,31 lyophilized, and stored as 500-μg aliquots at −20°C. Contamination by other proteins was excluded by analytical isoelectric focusing and subsequent immunoblotting. The molar ratio of apoA-I and apoA-II to DMPC was 1:150. Before incubation with cells, proteoliposomes were dialyzed overnight at room temperature against DMEM.
Assay for PC-Derived Radiolabeled PA and DG Formation
Fibroblasts were seeded into 60-mm dishes; grown to 70% to 80% confluence in 4 mL of DMEM containing 10% FCS, 2 mmol/L l-glutamine, and 1% antibiotic/antimycotic solution (Sigma); and loaded with cholesterol. After 48 hours the medium was removed, and the cells were washed three times with PBS and incubated for 2 hours in 2 mL of DMEM-HEPES containing 0.05 μCi/mL l-lyso-3-phosphatidylcholine and [1-14C]palmitoyl (56 mCi/mmol). Thereafter, cells were washed five times in PBS and incubated for an additional 3 hours in 2 mL of DMEM-HEPES. At different times before stopping the 3-hour incubation, cells were stimulated with HDL3, proteoliposomes, or PBS (control value). To terminate the incubation the medium was removed, and the cell dishes were quickly placed in a liquid nitrogen bath. The cells were scraped from the dishes with a rubber policeman, once with 2 mL of ice-cold 0.2-mol/L NaOH and once with 2 mL of ice-cold distilled water. Radioactive lipids were extracted by the method of Folch et al.32 For transphosphatidylation experiments, the 3-hour incubation phase was performed in the presence of 0.3% butanol. To monitor PI-derived DG and PA, the final 24-hour period of cholesterol loading was performed in the presence of [14C]AA (0.5 μCi/mL). Thereafter, the cells were washed five times in PBS before starting the 3-hour incubation.
Separation of Radiolabeled PA, DG, and PBut by TLC
In most experiments, a double one-dimensional TLC (as described in Reference 3333 , with minor modifications) was used to separate phospholipids and neutral lipids of interest. In this approach, a series of samples were spotted 120 mm from the bottom of the plate. To resolve labeled neutral lipids (DG, fatty acids, and triglycerides) from phospholipids that remained at the origin, the plates were developed twice in toluol/ether/ethanol/triethylamine (100:80:4:2, vol/vol). After the first run to 200 mm, the plates were thoroughly dried and developed a second time with the same solution to 160 mm. Plates were then cut 8 mm above the origin (ie, 128 mm above the lower edge of the plate), rotated 180°, and developed to the top with chloroform/methanol/ammonium hydroxide (65:35:5, vol/vol). After the plates had dried thoroughly, autoradiography was performed by using Kodak X-OMAT film for 7 to 14 days. Radioactive bands were cut from the silica plates, placed in scintillation vials containing 10 mL Ultima-Gold scintillation fluid, and quantitated by liquid scintillation counting in a 1214 scintillation counter (LKB). The identities of labeled bands were determined based on RF values obtained for authentic neutral lipids and phospholipids (from Sigma) visualized by iodine staining. PBut standard was prepared as described in Reference 3434 .
Measurement of Choline Release
Fibroblasts were seeded into 35-mm dishes and grown to confluence in 4 mL of DMEM containing 10% FCS, 2 mmol/L l-glutamine, and 1% antibiotic/antimycotic solution (Sigma). Cells were labeled for 48 hours in choline-free DMEM containing [methyl-14C]choline (2 μCi/mL), 10% FCS, 2 mmol/L l-glutamine, and 1% antibiotic/antimycotic solution. During the labeling interval 30±9% of the [14C]choline was taken up by the cells. After 48 hours the medium was removed, and the cells were washed three times with PBS and incubated for 20 minutes in 2 mL of DMEM-HEPES containing unlabeled choline (1 mmol/L). Cells were again washed in PBS and incubated for another 20 minutes in fresh choline-free DMEM-HEPES. The medium was then removed, and the cells were washed again twice in 1 mL of PBS and incubated in 1 mL of choline-free DMEM-HEPES for an additional 2 hours. At different times before stopping the 2-hour incubation, cells were stimulated with either HDL3 (25 μg/mL) or PBS. To terminate the incubation, the medium was removed and transferred to tubes prefilled with 4 mL ice-cold chloroform/methanol (1:2, vol/vol). After removing the medium, the cell dishes were rapidly immersed in a liquid nitrogen bath and stored at −20°C until further preparation. The cells were scraped off the dishes with a rubber policeman, once with 1 mL of ice-cold 0.2-mol/L NaOH and once with 1 mL of ice-cold distilled water. The aqueous and lipid phases of the medium and the cell extracts were separated as described by Folch et al.32 The aqueous phase of the chloroform/methanol extract was dried and resuspended in 50% ethanol.
Measurement of the Incorporation of [Methyl-14C]Choline Into Cell Lipids
Cholesterol-loaded cells were incubated for 30 minutes in choline-free DMEM (2 mL), washed twice with PBS, and then incubated in DMEM (without choline) plus [methyl-14C]choline (0.5 μCi/mL) for 45 minutes. HDL3 or PBS (for control) was added 30 minutes before terminating the reaction. Aliquots of the medium were directly counted for radioactivity. The cellular lipids were extracted and then separated by double one-dimensional TLC, as described above. The band corresponding to [14C]PC was cut out and counted for radioactivity.
Separation of [14C]Choline-Labeled Metabolites
Aqueous [14C]choline-labeled metabolites were separated and quantitated by a procedure modified from that described by Yavin.35 Nonradioactive choline and phosphocholine (50 μg of each) were added to an aliquot of the aqueous phase, which amounted to one half of the total. The samples with standards added were dried by vacuum centrifugation. The residue was dissolved in 75 μL of 50% ethanol, spotted onto silica gel 60 plates, and developed to the top in a solvent composed of 0.6% NaCl/methanol/35% NH3 (100:100:3, vol/vol). The plates were air dried for 30 minutes, and the standards were visualized by exposure of the plates to iodine vapor. Iodine was sublimated, and spots corresponding to Cho (RF 0.09) and ChoP (RF 0.39) were cut out, eluted for 48 hours with 10 mL of Ultima-Gold scintillation fluid, and then counted for radioactivity by liquid scintillation in a 1214 scintillation counter.
Cholesterol-loaded fibroblasts were washed three times with PBS and then incubated for 3 hours at 37°C in phosphate-free DMEM containing 20 μCi/mL of [32P]orthophosphoric acid and 20 mmol/L HEPES. Cells were stimulated with HDL3 or proteoliposomes, and after the indicated time the reaction was terminated by removing the medium and scraping the cells off the dishes with ice-cold 0.7% NaCl by use of a rubber policeman. Extraction was performed with chloroform/methanol/HCl (100:200:2, vol/vol). [32P]phospholipids were developed in chloroform/methanol/acetone/acetic acid/water (60:20:23:18:12, vol/vol) with potassium oxalate–impregnated silica 60 plates (Merck). Bands corresponding to PIP and PIP2 were cut out from the silica plates, placed in scintillation vials containing 10 mL Ultima-Gold scintillation fluid, and quantitated by liquid scintillation counting in a 1214 scintillation counter. The identities of labeled bands were determined based on RF values obtained for authentic phospholipids (from Sigma) visualized by iodine staining, as described in Reference 3636 .
Protein concentrations were measured according to the method of Bradford,37 using bovine serum albumin as the standard. Each experiment was performed in triplicate and repeated three to eight times, as indicated in appropriate legends. Unless otherwise indicated, data represent the mean±SD of data obtained from a representative experiment.
Time Course of HDL3-Stimulated PC-Derived DG and PA Production
To characterize HDL3-mediated PC breakdown, we examined the time course of DG and PA levels after addition of HDL3. Cholesterol-loaded human skin fibroblasts were labeled with [14C]lyso-PC for 2 hours at 37°C. Under these conditions, ≈70% of the radiolabel incorporated into phospholipids was acylated to PC, without detectable labeling of PI (data not shown). Incubation of the cultures with low concentrations of HDL3 (20 μg/mL) resulted in a biphasic accumulation of [14C]PA, as shown in Fig 1a⇓. The early phase was rapid, peaking at 30 to 60 seconds up to 140% of control values (P<.001), followed by a rapid decline observed between 2 and 4 minutes after stimulation (4 minutes: P<.02 versus 30 seconds after HDL3 stimulation). The second phase of HDL3-stimulated [14C]PA generation reached a maximum between 7 and 10 minutes, decayed slowly thereafter, and was sustained above control values for at least 2 hours. In contrast to the first PA response, this second increase of PA varied to some extent from one experiment to another (maximal increase: 1.5 to 3-fold).
HDL3-induced PA production preceded DG production. A significant increase in [14C]DG was observed 2 minutes after stimulation (P<.05), rising up to 140% of control values until 4 minutes (Fig 1b⇑). In all experiments the rise of [14C]DG was detectable for at least 2 hours.
Effect of HDL3 Concentration on PC-Derived DG and PA Production
As shown in Fig 2a⇓ and 2b⇓, the early HDL3-mediated PC breakdown in cholesterol-loaded fibroblasts (PA, 2 minutes; DG, 4 minutes) occurred in a concentration range of 10 to 75 μg/mL. It was dose dependent up to 20 μg/mL. Maximum DG and PA increases were observed at a concentration of 20 μg/mL. However, this early PC breakdown was not detectable at HDL3 concentrations above 75 μg/mL. At 30 minutes, by contrast (Fig 2c⇓ and 2d⇓), DG and PA increases were visible at a wide concentration range of HDL3 of between 5 and 75 μg/mL, with two peak maxima at 50 μg/mL and 500 to 750 μg/mL.
Time Course of HDL3-Stimulated Choline and Phosphocholine Production
The observations that the PA production preceded the DG production and that the initial DG increase coincided with a temporary PA decrease suggested that PA was formed via the PLD pathway and that DG was at least partly derived from PA by the consecutive action of PLD and PA-PPH. To exclude the possibility that PA was derived from a rapidly phosphorylated pool of DG, we examined the effect of HDL3 on water-soluble choline metabolite levels in cholesterol-loaded human fibroblasts labeled with [methyl-14C]choline. If PA arises through the action of PLD, Cho production should precede, or at least coincide with, the rise in ChoP.
After a 48-hour labeling period with 2.0 μCi/mL [methyl-14C]choline, followed by a 40-minute chase incubation with unlabeled choline and a 2-hour incubation in choline-free DMEM, about 80% of the radioactivity was found in PC. More than 90% of the water-soluble radiolabeled choline metabolites consisted of [14C]ChoP and [14C]Cho. About 3% of the total [14C]ChoP pool and 4% of the total [14C]Cho pool were detected in the incubation medium.
Addition of 20 μg/mL HDL3 for 30 to 60 seconds before the end of the 2-hour incubation period in choline-free DMEM significantly elevated intracellular [14C]Cho levels (Fig 3a⇓; 30 seconds: P<.02; 60 seconds: P<.001). Maximal elevation up to 160% of control values was observed if HDL3 was added 6 minutes before stopping the reaction. The rise in [14C]ChoP occurred later (Fig 3b⇓) and was significant after 2 minutes of stimulation (P<.001), and levels thereafter decreased to ≈6% below initial values for 20 minutes, followed by a sustained increase for at least 2 hours. Extracellularly, the [14C]ChoP level also significantly increased after 2 minutes of stimulation (P<.001), peaked at 5 minutes, and remained elevated for at least 2 hours (Fig 3d⇓). These data demonstrate that HDL3-mediated formation of intracellular Cho preceded that of both intracellular and extracellular ChoP. In contrast to intracellular [14C]Cho, extracellular [14C]Cho rapidly decreased to 60% of initial levels after 2 minutes and remained below basal levels for 2 hours (Fig 3c⇓). As choline can permeate the cell membrane, the most likely explanation for the observed discrepancies of intracellular and extracellular [14C]Cho levels was the concomitant HDL3-mediated activation of PC breakdown and PC resynthesis. To investigate whether HDL3 stimulates PC synthesis, its effect on [methyl-14C]choline incorporation uptake into cellular lipids was tested. The results are shown in Table 1⇓. Indeed, HDL3 (20 μg/mL and 200 μg/mL) stimulated PC synthesis when cells were supplied with exogenous choline. Radioactivity associated with Cho in the supernatant decreased to 6% to 8% of control values under the influence of HDL3. Concomitantly, PC resynthesis was enhanced threefold to fourfold. HDL3, therefore, stimulated both the breakdown and the resynthesis of PC.
Phosphatidylbutanol Accumulation in Response to HDL3
PLD hydrolyzes phospholipids to yield the free polar headgroup Cho and PA. In the presence of primary alcohols, however, PLD catalyzes a phosphatidyl-transfer reaction, producing phosphatidylalcohols.38 Since phosphatidylalcohols are not attacked by PA-PPH and the production of phosphatidylalcohol is mediated exclusively by PLD, synthesis of phosphatidylalcohols is a specific and unequivocal marker for involvement of this enzyme.39 40 To confirm activation of PLD, we examined the ability of HDL3-stimulated [14C]lyso-PC–labeled fibroblasts to form [14C]PBut in the presence of 0.3% butanol.
Maximum [14C]PBut formation was attained at 0.3% vol/vol butanol. This concentration did not result in cell lysis and did not influence cell growth as measured over a period of 2 days, indicating that metabolic responses were measured under quasiphysiological conditions (data not shown). Stimulation of cholesterol-loaded fibroblasts with 20 μg/mL HDL3 in the presence of 0.3% butanol resulted in the time-dependent accumulation of [14C]PBut, as shown in Fig 4⇓. Significant increases of [14C]PBut above control values were apparent at 30 seconds (P<.05), and levels rose rapidly until 6 minutes and more slowly until 2 hours. [14C]PBut levels remained unchanged in the absence of stimulation (PBS).
Effect of Butanol, Propranolol, and PLC Inhibitors on HDL3-Induced DG Levels
Since HDL3-mediated PBut, Cho, and PA formation occurred prior to the sustained increase in DG, it seemed likely that PA was being converted to DG by the action of PPH. To assess this assumption, butanol was used to “trap” PA moieties such as PBut, which is not attacked by PPH. The results are presented in Fig 5⇓. The major part of DG formation at short incubation times (t<15 minutes) was inhibited in the presence of 0.3% butanol, suggesting a major role for PA in contributing to HDL3-mediated DG formation. However, a minor part of the HDL3-induced DG increase still occurred in the presence of butanol over the entire incubation time. In particular, both a transient increase at 2 to 4 minutes and a sustained increase in DG from 15 minutes onward were detectable despite the presence of butanol. These early and late DG increases coincided with the ChoP increases shown in Fig 3⇑, indicating the additional activation of PC-specific PLC at these incubation times.
As shown in Table 2⇓, propranolol (100 μmol/L), a direct inhibitor of PPH,26 diminished the HDL3-induced DG increases by more than 50% at 4 minutes and 10 minutes (20 μg/mL HDL3), further suggesting a major role for PA in contributing to DG formation. However, at longer incubation times (30 minutes) or higher concentrations of HDL3 (500 μg/mL), the DG response was inhibited to a lesser degree by this treatment. HDL3 concentrations above 75 μg/mL did not produce a significant DG increase at 4 minutes, irrespective of the presence of propranolol.
To further evaluate the contribution of PI- or PC-specific PLC, we studied the influence of the PC-specific PLC inhibitor D609 (tricyclodithiocarbonate; 30 μg/mL) and the PI-specific PLC inhibitor U 73122 (15 μmol/L) on the HDL3-induced PC hydrolysis. As shown in Table 3⇓, the PC-specific PLC inhibitor slightly reduced the DG increase at 4 minutes but markedly reduced the PC-derived DG increase at 30 minutes. In the presence of the PI-specific PLC inhibitor U 73122, both the basal DG level and the HDL3-induced DG increase at 30 minutes were decreased. At 2 minutes, by contrast, the HDL3-induced accumulation of [14C]DG was enhanced in the presence of the inhibitor.
Altogether, these data indicate (1) that the major part of the PC-derived DG increase at t<15 minutes is produced by a coupled PLD/PPH pathway and (2) that PC-specific PLC is predominantly involved in the sustained HDL3-mediated PC-derived DG accumulation.
Involvement of Specific HDL Receptor Binding Sites
HDL particles treated with proteolytic enzymes or TNM to digest or modify apoproteins do not bind to the HDL receptor and do not promote translocation of cholesterol from intracellular pools to the cell surface.4 29 We therefore examined the influence of TNM modification of HDL3 on PC breakdown. The results are shown in Fig 6⇓. A significant DG or PA increase in cholesterol-loaded fibroblasts in response to HDL3-TNM (20 μg/mL) was not detected at short incubation times (2 or 4 minutes), suggesting that HDL3-mediated early PC breakdown involves the interaction of HDL apoproteins with the cell surface. DG and PA increases 30 minutes after stimulation were only partially inhibited by TNM modification of HDL3. We therefore examined the possibility that a lipid component of HDL3 induces a sustained PC hydrolysis. However, protein-free phospholipid vesicles with different phospholipids (DMPC, dipalmitoylphosphatidylcholine, egg-lecithin) did not induce a significant increase of DG or PA during the examined time course (data not shown).
To examine whether or not HDL3 itself has a PLC or PLD activity on pure phospholipid substrates, we incubated HDL3 (500 μg/mL) in cell-free medium with phospholipid particles containing radiolabeled PC. As demonstrated in Table 4⇓, the radioactivity associated with the DG and PA position on the TLC plates did not change during a 24-hour incubation at 37°C in DMEM-HEPES either in the presence or in the absence of HDL3.
Effect of PKC Depletion on HDL3-Stimulated [14C]PA and [14C]DG Formation
A main downstream effector of phospholipid-mediated second-messenger responses is PKC, which in its active form phosphorylates intracellular proteins. On the other hand, phospholipid breakdown can also be modulated by PKC, eg, in the sense of a positive or negative feedback mechanism. To study the influence of PKC on PC-specific PLD, cholesterol-loaded fibroblasts were exposed to 200 nmol/L of PMA for 24 hours. Under this condition, virtually complete downregulation of PKC occurs.27 The effects of PKC depletion were analyzed by comparing [14C]PA and [14C]DG levels in cholesterol-loaded fibroblasts preincubated with PMA or the inactive phorbol ester 4-α-phorbol-12,13-didecanoate (control value). The results of a typical experiment are shown in Fig 7⇓. [14C]DG formation in response to 20 μg/mL HDL3 was enhanced at 2.5 minutes but was diminished at longer incubation times (30 minutes) by downregulation of PKC. [14C]PA formation, by contrast, was not influenced or was slightly inhibited at 2.5 minutes and 7.5 minutes but was markedly enhanced at 30 minutes. These data demonstrated that PC-PLC is under negative control of PKC at short incubation times (2.5 minutes) and under positive control of PKC at longer incubation times (30 minutes). Conversely, PC-PLD is apparently not significantly influenced by PKC at short incubation times (2.5 minutes or 7.5 minutes) and under a negative control of PKC at longer incubation times.
HDL3-Mediated PI Breakdown
In human fibroblasts, a variety of agonists, including many growth factors, activate PI-specific PLC, resulting in the breakdown and subsequent resynthesis of phosphoinositides and the generation of DG, which is thereafter converted to PA. To assess whether HDL3 acts via this pathway, we examined the effect of HDL3 on PI metabolism by [32P]orthophosphoric acid labeling. As shown in Fig 8⇓, addition of HDL3 (20 μg/mL or 200 μg/mL) produced a rapid decrease in endogenous PIP2. 32P-labeled PIP2 fell to 42% and 58% of the initial values at 30 seconds after stimulation with 20 and 200 μg/mL HDL3, respectively. Thereafter, the levels rose and reached 95% and 72% of those in unstimulated cells after 3 minutes. Concomitantly, PIP levels showed a transient decrease to 35% and 70% of unstimulated levels at 15 to 30 seconds under the influence of 20 and 200 μg/mL HDL3, respectively. HDL3-mediated PI turnover was completely abolished in the presence of the PI-PLC inhibitor U 73122 (15 μmol/L).
To confirm HDL3-mediated PI hydrolysis, cholesterol-loaded fibroblasts were prelabeled with [14C]AA, which is known to be incorporated mainly into PI. Cholesterol-loaded human skin fibroblasts were labeled with [14C]AA (2 μCi/mL) for 48 hours at 37°C. During that time approximately 85% of the radiolabel was found in PI (data not shown). Incubation of the cultures with HDL3 (100 μg/mL) resulted in a rapid monophasic [14C]DG increase that peaked at 30 to 60 seconds and reached nearly basal levels after 5 minutes. The DG increase was followed by an increase in [14C]PA, with a peak maximum at 4 minutes, as shown in Fig 9⇓. [14C]PA reached basal levels after 10 minutes.
Stimulation of PC-Derived PA and DG Accumulation by ApoA-I and ApoA-II Proteoliposomes
To examine the involvement of apoA-I in the HDL3-induced second-messenger response, apoA-I proteoliposomes were tested for their ability to induce PC turnover in human skin fibroblasts. As shown in Fig 10⇓, exposure of cells to apoA-I proteoliposomes (10 μg/mL) resulted in a rapid PA increase to 150% of basal levels, with a peak maximum at 30 seconds, followed by an increase in DG to 150% of baseline. At 2 minutes, when PA had reached basal levels again, the DG increase was maximal. DG returned to almost basal levels after 7 minutes. The kinetics and the extent of PA and DG increase induced by apoA-I proteoliposomes resemble the first phase observed with 20 μg/mL HDL3, as shown in Fig 1⇑. As the PA response preceded the DG increase, it appears that DG formation occurs at least partly by a coupled PLD/PPH pathway. Activation of PLD was confirmed by the detection of [14C]PBut in the presence of 0.3% butanol (Fig 11⇓, inset).
ApoA-II proteoliposomes induced an increase of PA, PBut, and DG similar to apoA-I proteoliposomes (Fig 11⇑). However, the maximal increases of [14C]PA and [14C]DG were twofold to threefold higher than with apoA-I proteoliposomes and were reached at lower concentrations (apoA-I: 5 to 10 μg/mL; apoA-II: <5 μg/mL). DMPC liposomes alone did not induce substantial PC hydrolysis. The formation of [14C]PA and (to a lesser extent) [14C]DG was suppressed at higher concentrations of proteoliposomes. The inhibition of PC breakdown at higher proteoliposome concentrations was more pronounced with apoA-II than with apoA-I.
These data suggest that both apolipoprotein A-I and apolipoprotein A-II can mimic the HDL3-mediated PA response at 30 seconds and the DG increase at between 2 and 4 minutes.
In contrast to HDL3, both apoA-I and apoA-II proteoliposomes (10 μg/mL) failed to cause a depletion of PIP2 or PIP during a 4-minute time course (Fig 8⇑). In some experiments, however, the addition of apoA-I or apoA-II proteoliposomes caused a slight increase of PIP2 and PIP at between 0 and 60 seconds (apoA-I: in 6 of 8 experiments; apoA-II: in 4 of 6 experiments, as shown in Fig 8⇑).
In general, the increase of PIP2 and PIP is a compensatory response to the PLC-mediated breakdown of PIP2 and other phosphoinositides to replenish pools of depleted inositol phospholipids. However, such a mechanism does not appear to account for the effect of apoA-I and apoA-II proteoliposomes. Since phosphoinositides are derived from cytidine-diphosphate-DG formed from PA, the PIP2 and PIP synthesis might be due to the rapid activation of PLD induced by apoA-I and apoA-II. On the other hand, we cannot entirely exclude the possibility that a modest activation of PI-specific PLC induced by apoA-I and apoA-II may be compensated for by increased phosphoinositide synthesis. Other instances in which PC breakdown occurs without PI turnover were found in T lymphocytes exposed to interleukin 1,41 in epidermal growth factor–stimulated Swiss 3T3 fibroblasts,42 and also in adipose cells exposed to HDL3.24 To exclude the possibility that supplying membranes with PC, through liposome-cell interactions, can inhibit PI-specific PLC, we examined the HDL3-induced PI turnover in the presence of 10 μg/mL apoA-I proteoliposomes. We could not detect a difference of HDL3-mediated PI turnover in the presence or absence of proteoliposomes in four independent series of experiments (data not shown).
Influence of Cholesterol Loading on the HDL3-Induced Second-Messenger Response
To examine whether the HDL3-induced second-messenger response is influenced by cholesterol loading, we compared the accumulation of [14C]PA and [14C]DG in cholesterol-loaded and control cells. As shown in Fig 12⇓, the basal value of [14C]PA was increased by cholesterol loading. [14C]DG was not significantly influenced by this treatment. The HDL3-induced PA increase at 2 minutes was markedly decreased in cells that had not been loaded with cholesterol. The HDL3-induced DG increase, by contrast, was increased in these cells. The HDL3-induced PC hydrolysis at longer incubation times (15 minutes) was similar in cholesterol-loaded and nonloaded cells, apart from a slight relative increase of PA in nonloaded cells due to a decreased basal value in this cell type. The HDL3-induced PI turnover was not influenced by cholesterol loading (data not shown).
Our results unequivocally demonstrate that both PI-specific PLC and PC-specific PLC as well as PC-specific PLD are activated upon incubation of HDL3 with cholesterol-loaded fibroblasts. HDL3 stimulates both the breakdown and the resynthesis of PC and PI.
In lyso-PC–labeled cholesterol-loaded human fibroblasts, addition of HDL3 stimulated a biphasic accumulation of [14C]PA, with peak maxima at 30 to 60 seconds and 7 to 10 minutes. The rapid increase and the similar kinetics for the release of the two direct products of PLD, [14C]PA (in [14C]lyso-PC–labeled cells) and [14C]Cho (in [methyl-14C]choline–labeled cells), suggested that HDL3 stimulates activation of PLD. PLD activation was confirmed by the detection of HDL3-mediated phosphatidyl transfer, producing [14C]PBut instead of [14C]PA in the presence of butanol. That [14C]PA, [14C]Cho, and [14C]PBut formation preceded [14C]DG formation is consistent with the conclusion that the [14C]PA produced by PLD serves as a precursor for [14C]DG. By comparing the DG kinetics in the absence and presence of butanol or propranolol (which is a direct inhibitor of PPH), we could demonstrate that the PC-derived DG formation at t<15 minutes primarily occurred by a coupled PLD/PPH pathway and that a major part of the sustained DG formation was directly derived from PC by means of a PC-PLC. Activation of PC-specific PLC was also demonstrated by means of the PC-PLC inhibitor D609. Moreover, we could demonstrate that the DG increase was paralleled by an increase of cellular ChoP, the direct water-soluble product of PC-PLC.
In addition to PC-PLC and PC-PLD, HDL3 stimulated PI-specific PLC. PI turnover was rapid and brief. A decrease in PIP2 and PIP and an increase in PI-derived DG were detected 30 seconds after stimulation. The PI-derived DG increase in [14C]AA-labeled cells preceded the PI-derived PA increase, indicating the activation of PI-specific PLC and the subsequent activation of DG kinase. On the basis of the observation of IP3 formation and calcium transients, HDL3-induced activation of PI-PLC was previously reported by Bochkov et al.21 In the present study we provide direct evidence on the level of PI breakdown and DG formation.
It is conceivable that the activities of PI-PLC and PC-PLC are closely related. Stimulation of both enzymes was observed in a similar HDL3 concentration range. Moreover, pretreatment of cells with the PI-PLC inhibitor U 73122 inhibited the formation of both PI- and PC-derived DG. Finally, depletion of PKC reduced the formation of PC-derived DG at longer incubation times. It is therefore possible that the HDL3-induced activation of PI-PLC results in the stimulation of PKC, which in turn activates PC-PLC. PKC-mediated PC-PLC activation has been observed in other cell systems, such as insulin-stimulated hepatocytes43 and bradykinin-treated fibroblasts.44 The HDL3-mediated activation of PKC was previously reported by Mendez et al8 in fibroblasts and by us45 in platelets. In contrast to DG, sustained HDL3-mediated PA formation was greatly enhanced by downregulation of PKC, consistent with a PKC-dependent PC-PLD desensitization limiting PA formation at longer incubation times. PC-PLD desensitization has previously been reported in other cell types.46 47 Recently it was proposed that PKC may play a crucial role in PLD desensitization.48
Interestingly, the early activation of PC-PLD at 0.5 to 3 minutes was not detectable at HDL3 concentrations >75 mg/mL. The rapid activation of PI-PLC and the sustained PC hydrolysis, by contrast, were observed in a wide concentration range up to 750 μg/mL. One possibility for this phenomenon was that a desensitization effect exerted by PKC over PLD also suppresses the early PLD activity. However, the early PA formation was not influenced or was slightly inhibited by PKC downregulation, indicating that this second-messenger response is not under the negative control of PKC.
The unique character of the early PLD activation was further strengthened by the observation that only this second-messenger response was enhanced by cholesterol loading. Moreover, only the early activation of PLD, but not the sustained PC hydrolysis or the rapid PI hydrolysis, could be mimicked by apoA-I and apoA-II proteoliposomes. Hence, it cannot be entirely excluded that early activation of PLD may occur independently of PI hydrolysis and sustained PC hydrolysis. If this is the case, the signal coming directly from the putative HDL/apoA-I receptor to PC-PLD could be attenuated by receptor modification or receptor downregulation.49
HDL3 is a complex particle with various lipid and protein components, many of which might conceivably be involved in triggering cell signaling. To address this issue in more detail, we examined the HDL3-induced second-messenger response after modification of HDL3 with TNM, a protein cross-linking reagent. TNM modification has been reported to inhibit specific effects of HDL3 in human skin fibroblasts, such as binding,5 activation of PKC,8 and cholesterol mobilization.4 7 We found that TNM treatments considerably reduced HDL3-mediated cell signaling. Moreover, cell signaling could not be induced by protein-free phospholipid vesicles, and we could not detect PLC or PLD activity associated with HDL3. Therefore, we conclude that protein components are the most likely candidates for all of the responses examined and that the HDL3-mediated activation of cellular phospholipases is mediated by specific binding sites.
Specific HDL binding sites with nonidentical characteristics have been described by many groups (reviewed in Reference 33 ). It is therefore possible that HDL interacts with functionally different receptors and induces different cellular signals. Two specific metabolic responses of HDL are known: the mobilization of intracellular cholesterol4 5 6 7 8 9 and the induction of mitogenic effects.10 11 12 13 14 15 16 17 18 In previous studies, HDL3 receptor-coupled cholesterol efflux was demonstrated by labeling cells with the cholesterol precursor mevalonolactone and monitoring the transport of newly synthesized cholesterol to the cell membrane.4 7 8 9 In human skin fibroblasts, saturation of specific HDL3 binding was observed at HDL3 concentrations of ≈20 μg/mL.5 The maximal response of early PLD activation was observed at a similar concentration. Moreover, artificial apoA-I proteoliposomes induce cholesterol mobilization effectively,50 51 and early PLD activation described in this study was inducible by cholesterol loading of the cells. Therefore, early PLD activation, which could be mimicked by apolipoprotein proteoliposomes, is the likely mechanism promoting cholesterol mobilization.
Other studies described a strong mitogenic effect of HDL in vascular endothelial cells and smooth muscle cells that increased up to 500 μg/mL HDL.10 12 Increases in DG, inositol phosphates, and PA have been implicated as important mediators of mitogenesis.52 53 The addition of growth factors to quiescent cell cultures stimulates increases in one or more of these metabolites. Since in many cases the stimulation of DNA synthesis requires the continued presence of the mitogen for several hours, it is likely that prolonged second-messenger generation such as PC-derived DG and PA may play an important role in the transduction of mitogenic signals.20 54 55 Mostly, mitogen-induced hydrolysis of PI is rapid and transient, whereas PC hydrolysis is sustained.19 55 56 Thus, it is possible that rapid PI turnover and sustained PC-PLC activation are induced by a growth factor–like activity on the HDL particle. Activation of PI-PLC, PC-PLC, as shown here, and the previously described mitogenic effect of HDL could be observed over a wide concentration range of HDL. Moreover, HDL3-mediated PC hydrolysis is accompanied by PC resynthesis, which is thought to provide regulation of DG levels in mitogenesis. However, there are also arguments supporting involvement of apolipoproteins in mitogenesis. ApoA-I was reported to substitute for HDL in promoting the growth of SV-40–transformed REF52 fibroblasts.57 In addition, both HDL3 and apoA-I proteoliposomes were able to stimulate DNA synthesis in the adenocarcinoma cell line A549.18
In conclusion, we have shown that multiple phospholipases are activated upon stimulation with HDL3. Our results demonstrate that HDL3-mediated cell signaling is complex and that more than one agonist activity may be involved. Clearly, further studies are necessary to evaluate the exact biologic roles of the HDL3-induced second-messenger molecules described in this study.
Selected Abbreviations and Acronyms
|DMEM||=||Dulbecco’s modified Eagle’s medium|
|FCS||=||fetal calf serum|
|PKC||=||protein kinase C|
|PMA||=||phorbol 12-myristate 13-acetate|
|PPH||=||phosphatidic acid phosphohydrolase|
We thank Dr Paul Cullen for his helpful comments in the preparation of this manuscript. The manuscript contains part of an MD thesis work (Holger Reinecke). The work was supported by a fellowship from the Deutscher Akademischer Austauschdienst (D.A.A.D.) to Jerzy-Roch Nofer.
Rothblat GH, Phillips MC. Mechanism of cholesterol efflux from cells: effects of acceptor structure and concentration. J Biol Chem. 1982;257:4775-4782.
Slotte JP, Oram JF, Bierman EL. Binding of high density lipoproteins to cell receptors promotes translocation of cholesterol from intracellular membranes to the cell surface. J Biol Chem. 1987;262:12904-12907.
Oram JF, Brinton EA, Bierman EL. Regulation of high density lipoprotein receptor activity in cultured human skin fibroblasts and human arterial smooth muscle cells. J Clin Invest. 1983;72:1611-1621.
Oram JF, Mendez AJ, Slotte JP, Johnson TF. High density lipoproteins mediate removal of sterol from intracellular pools but not from plasma membranes of cholesterol loaded fibroblasts. Arterioscler Thromb. 1991;11:403-414.
Mendez AJ, Oram JF, Bierman EL. Protein kinase C as a mediator of high density lipoprotein dependent efflux of intracellular cholesterol. J Biol Chem. 1991;266:10104-10111.
Gospodarowicz D, Hirabayashi K, Giguère L, Tauber J-P. Factors controlling the proliferative rate, final cell density, and life span of bovine vascular smooth muscle cells in culture. J Cell Biol. 1981;89:568-578.
Björkerud S, Björkerud B. Lipoproteins are major and primary mitogens and growth promoters for human arterial smooth muscle cells and lung fibroblasts in vitro. Arterioscler Thromb. 1994;14:288-298.
Boyd NF, McGuire V. Evidence of association between plasma high-density lipoprotein cholesterol and risk factors for breast cancer. J Natl Cancer Inst. 1990;82:460-468.
Favre G, Tazi KA, Le Gaillard F, Bennis F, Hachem H, Soula G. High density lipoprotein 3 binding sites are related to DNA biosynthesis in the adenoma cell line A549. J Lipid Res. 1993;34:1093-1106.
Pörn MI, Åkerman KEO, Slotte JP. High density lipoproteins induce a rapid and transient release of Ca2+ in cultured fibroblasts. Biochem J. 1991;279:29-33.
Milligan G. An indirect pathway of receptor-mediated 1,2-diacylglycerol formation in mast cells. Biochem J. 1988;255:1-13.
Havel RJ, Eder HA, Bragdon JH. The distribution and chemical composition of ultracentrifugally separated lipoproteins in human serum. J Clin Invest. 1955;34:1345-1353.
Brinton EA, Oram JF, Chen C-H, Albers JJ, Bierman EL. Binding of high density lipoprotein to cultured fibroblasts after chemical alteration of apoprotein amino acid residues. J Biol Chem. 1986;261:495-503.
Chen C, Albers JJ. Characterization of proteoliposomes containing apoprotein A-I: a new substrate for the measurement of lecithin:cholesterol acyl transferase activity. J Lipid Res. 1982;23:680-691.
von Eckardstein A, Funke H, Walter M, Altland K, Benninghoven A, Assmann G. Structural analysis of human apolipoprotein A-I variants: amino acid substitutions are nonrandomly distributed throughout the apolipoprotein A-I primary structure. J Biol Chem. 1990;265:8610-8617.
Folch J, Lees M, Sloane-Stanley GH. A simple method for the isolation and purification of total lipids from animal tissues. J Biol Chem. 1957;26:497-509.
Gruchalla RS, Dinh TT, Kennerly DA. An indirect pathway of receptor-mediated 1,2-diacylglycerol formation in mast cells. J Immunol. 1990;144:2334-2342.
Kaszkin M, Seidler L, Kast R, Kinzel V. Epidermal-growth-factor-induced production of phosphatidylalcohols by HeLa cells and A431 cells through activation of phospholipase D. Biochem J. 1992;287:51-57.
Yavin E. Regulation of phospholipid metabolism in differentiating cells from rat brain cerebral hemispheres in culture. J Biol Chem. 1976;251:1392-1397.
Billah MM, Lapetina EG. Rapid decrease of phosphatidylinositol 4,5-bisphosphate in thrombin-stimulated platelets. J Biol Chem. 1982;257:12705-12708.
Pai J-K, Siegel MI, Egan RW, Billah MM. Phospholipase D catalyzes phospholipid metabolism in chemotactic peptide-stimulated HL-60 granulocytes. J Biol Chem. 1988;263:12472-12477.
Huang C, Cabot MC. Phorbol diesters stimulate the accumulation of phosphatidate, phosphatidylethanol, and diacylglycerol in three cell types. J Biol Chem. 1990;265:14858-14863.
Cook SJ, Wakelam MJO. Epidermal growth factor increases sn-1,2-diacylglycerol levels and activates phospholipase D-catalysed phosphatidylcholine breakdown in Swiss 3T3 cells in the absence of inositol-lipid hydrolysis. Biochem J. 1992;285:247-253.
van Blitterswijk WJ, Hilkmann H, de Widt J, van der Bend RL. Phospholipid metabolism in bradykinin-stimulated human fibroblasts. J Biol Chem. 1991;266:10344-10350.
Nofer J-R, Walter M, Seedorf U, Assmann G. HDL3 modulates platelet aggregation by activation of protein kinase C. Circulation. 1993;88(suppl I):I-457. Abstract.
McKenzie FR, Seuwen K, Pouysségur J. Stimulation of phosphatidylcholine breakdown by thrombin and carbachol but not by tyrosine kinase receptor ligands in cells transfected with M1 muscarinic receptors. J Biol Chem. 1992;267:22759-22769.
Moehren G, Gustavsson L, Hoek JB. Activation and desensitization of phospholipase D in intact rat hepatocytes. J Biol Chem. 1994;269:838-848.
Mahlberg FH, Rothblat GH. Cellular cholesterol efflux: role of cell membrane kinetic pools and interaction with apolipoproteins A-I, A-II, and C’s. J Biol Chem. 1992;267:4541-4550.
Rozengurt E. Early signals in the mitogenic response. Science. 1986;234:161-166.
Martinson EA, Goldstein D, Brown JH. Muscarinic receptor activation of phosphatidylcholine hydrolysis. J Biol Chem. 1989;264:14748-14754.
Wright TM, Shin HS, Raben DM. Sustained increase in 1,2-diacylglycerol precedes DNA synthesis in epidermal-growth-factor-stimulated fibroblasts. Biochem J. 1990;267:501-507.