Malonyl-CoA-independent Acute Control of Hepatic Carnitine Palmitoyltransferase I Activity

The mechanism of malonyl-CoA-independent acute control of hepatic carnitine palmitoyltransferase I (CPT-I) activity was investigated. In a first series of experiments, the possible involvement of the cytoskeleton in the control of CPT-I activity was studied. The results of these investigations can be summarized as follows. (i) Very mild treatment of permeabilized hepatocytes with trypsin produced around 50% stimulation of CPT-I activity. This effect was absent in cells that had been pretreated with okadaic acid (OA) and seemed to be due to the action of trypsin on cell component(s) distinct from CPT-I. (ii) Incubation of intact hepatocytes with 3,3′-iminodipropionitrile, a disruptor of intermediate filaments, increased CPT-I activity in a non-additive manner with respect to OA. Taxol, a stabilizer of the cytoskeleton, prevented the OA- and 3,3′-iminodipropionitrile-induced stimulation of CPT-I. (iii) CPT-I activity in isolated mitochondria was depressed in a dose-dependent fashion by the addition of a total cytoskeleton fraction and a cytokeratin-enriched cytoskeletal fraction, the latter being 3 times more potent than the former. In a second series of experiments, the possible link between Ca2+/calmodulin-dependent protein kinase II (Ca2+/CM-PKII) and the cytoskeleton was studied in the context of CPT-I regulation. The data of these experiments indicate that (i) purified Ca2+/CM-PKII activated CPT-I in permeabilized hepatocytes but not in isolated mitochondria, (ii) purified Ca2+/CM-PKII abrogated the inhibition of CPT-I of isolated mitochondria induced by a cytokeratin-enriched fraction, and (iii) the Ca2+/CM-PKII inhibitor KN-62 prevented the OA-induced phosphorylation of cytokeratins in intact hepatocytes. Results thus support a novel mechanism of short-term control of hepatic CPT-I activity which may rely on the cascade Ca2+/CM-PKII activation → cytokeratin phosphorylation → CPT-I de-inhibition.

Mitochondrial fatty acid oxidation in liver provides a major source of energy to this organ and supplies extrahepatic tissues with ketone bodies as a glucose-replacing fuel (1,2). Carnitine palmitoyltransferase I (CPT-I), 1 the carnitine palmitoyltransferase of the mitochondrial outer membrane, catalyzes the pace-setting step of long-chain fatty acid translocation into the mitochondrial matrix (1)(2)(3)(4)(5). Moreover, recent determination of flux control coefficients of the enzymes involved in hepatic long-chain fatty acid oxidation shows that CPT-I plays a pivotal role in controlling the flux through this pathway under different substrate concentrations and pathophysiological states (6,7). CPT-I is subject to long-term regulation in response to alterations in the nutritional and hormonal status of the animal (1, 2, 5). Short-term control of CPT-I activity involves inhibition by malonyl-CoA, the product of the reaction catalyzed by acetyl-CoA carboxylase (8). Since the latter enzyme is a key regulatory site of fatty acid synthesis de novo (cf. Refs. [1][2][3][4][5], malonyl-CoA inhibition of CPT-I allows an elegant explanation for the coordinate control of the partition of hepatic fatty acids into esterification and oxidation. As a matter of fact, evidence has accumulated during the last two decades highlighting the physiological importance of malonyl-CoA inhibition of CPT-I not only in liver but also in extrahepatic tissues (1,5).
During the last years, however, a novel mechanism of control of hepatic CPT-I activity has been put forward. Studies using permeabilized hepatocytes have shown that various agents exert short-term changes in CPT-I activity in parallel with changes in the rate of long-chain fatty acid oxidation (3,9). These short-term changes in hepatic CPT-I activity are assumed to be mediated by a malonyl-CoA-independent mechanism, since they survive cell permabilization, extensive washing of the permeabilized cells (to allow complete removal of malonyl-CoA), and subsequent preincubation of the cell ghosts at 37°C before determination of CPT-I activity (to allow recovery of the original conformational state of CPT-I) (10). Evidence has also been presented showing that the stimulation of hepatic CPT-I by the phosphatase inhibitor okadaic acid (OA), used as a model compound to study the short-term regulation of CPT-I, does not involve the direct phosphorylation of CPT-I (10). It has been recently shown that the OA-induced stimulation of CPT-I is prevented by KN-62, an inhibitor of Ca 2ϩ / calmodulin-dependent protein kinase II (Ca 2ϩ /CM-PKII) (11), and by taxol, a stabilizer of the cytoskeleton (12). These observations suggest that both activation of Ca 2ϩ /CM-PKII and disruption of the cytoskeleton may be necessary for the OA-induced stimulation of CPT-I to be demonstrated. It is conceivable that these two processes may be related, since Ca 2ϩ /CM-PKII is one of the protein kinases more actively involved in the control of the integrity of the cytoskeleton by phosphorylating cytoskeletal proteins (13). However, the events underlying this novel mechanism of control of CPT-I activity are as yet unknown. The present work was thus undertaken to study the molecular basis of the malonyl-CoAindependent short-term control of hepatic CPT-I activity.
Isolation and Incubation of Hepatocytes-Male Wistar rats (200 -250 g) which had free access to food and water were used in all experiments. Hepatocytes were isolated by the collagenase perfusion method and routinely incubated in Krebs-Henseleit bicarbonate buffer (pH 7.4) supplemented with 10 mM glucose and 1% (w/v) defatted and dialyzed bovine serum albumin as described before (9).
Assay of CPT-I Activity in Isolated Mitochondria-Mitochondria were isolated either from hepatocytes or from intact liver and CPT-I activity was measured as the malonyl-CoA-sensitive incorporation of radiolabeled L-carnitine into palmitoylcarnitine exactly as described before (10). When CPT-I activity was determined in suspensions of mitochondria containing cytoskeletal fractions (Figs. 3 and 4), CPT activity in the cytoskeletal fractions was subtracted from the CPT-I activity experimentally determined. In any event, CPT activity determined in those cytoskeletal fractions was always marginal and on the basis of protein content never accounted for more than 5% of the CPT-I activity measured in mitochondrial suspensions. Preparations of mitochondria were practically devoid of peroxisomes, as judged from the low recovery of catalase activity (Ͻ5%) in those preparations.
Assay of CPT-I Activity in Permeabilized Hepatocytes-After incubation of the hepatocytes with the additions indicated in each case, CPT-I activity was determined in digitonin-permeabilized hepatocytes as the tetradecylglycidate-sensitive incorporation of radiolabeled L-carnitine into palmitoylcarnitine (9,10). In the "one-step assay" (Table II), cell permeabilization and assay of enzyme activity are simultaneously performed (9). In other experiments, however, CPT-I activity was determined by a more complex procedure (Fig. 1). In this "two-step" assay, hepatocytes are permeabilized with digitonin, and then extensively washed with a medium containing 10 mM Tris-HCl (pH 7.4), 50 mM potassium fluoride, 100 mM KCl, 2.5 mM EDTA, and 2.5 mM EGTA. The permeabilized cell pellets were taken up in that medium with the additions indicated and CPT-I activity was subsequently determined after preincubation at 37°C for 5 min (10).
Western Blot Analysis of CPT-I-Mitochondrial fractions were subjected to SDS-PAGE using 10% polyacrylamide gels and proteins were further transferred onto nitrocellulose membranes. The blots were then blocked with 5% fat-free dried milk in phosphate-buffered saline supplemented with 0.1% Tween 20. They were subsequently incubated with the anti-CPT-I antibody (1:10,000) in phosphate-buffered saline/ Tween 20 for 2 h at 4°C, and washed thoroughly. The blots were then incubated with anti-sheep peroxidase-conjugated secondary antibody (1:10,000) for 1 h at room temperature, and finally subjected to luminography with an ECL detection kit.
Isolation of Cytoskeletal Fractions-Isolated hepatocytes were sedimented (2 min at 100 ϫ g) and resuspended in a cytoskeleton stabilizing buffer consisting of 10 mM Pipes (pH 6.8), 0.25 M sucrose, 3 mM MgCl 2 , 150 mM KCl, and 1 mM EGTA, supplemented with a proteinase/inhibitor mixture (11,15). The fraction corresponding to total cytoskeleton (Fraction I) as well as the fraction enriched in intermediate filaments (Fraction II) were prepared according to van Bergen en Henegouwen et al. (14).
Western Blot Analysis of Cytoskeletal Proteins-Cytoskeletal fractions were subjected to SDS-PAGE using 10% polyacrylamide gels and proteins were further transferred onto polyvinylidene fluoride membranes. The blots were then blocked with 2% Protifar (Nutricia, Zoetermeer, The Netherlands) in 50 mM Tris-HCl (pH 7.8), 100 mM NaCl, and 0.1% Tween 20 (TBST). They were subsequently incubated with the anti-actin, anti-tubulin, anti-cytokeratin 8, and anti-cytokeratin 18 monoclonal antibodies (1:10,000) in TBST with 0.2% Protifar for 1 h at 4°C, and washed thoroughly. The blots were incubated with peroxidase-conjugated secondary antibodies (1:10,000) for 1 h at room temperature, and finally subjected to luminography with the ECL detection kit.
Immunoprecipitation of 32 P-Labeled Cytokeratins-After isolation, hepatocytes were washed twice in phosphate-free Dulbecco's modified Eagle's medium supplemented with 1% (w/v) defatted and dialyzed bovine serum albumin. Hepatocytes (6 -8 mg of cellular protein in 1.5 ml of the aforementioned medium) were subsequently incubated in that medium at 37°C for 1 h with 0.2 mCi of [ 32 P]P i and subsequently exposed to the additions indicated. One ml of cells was rapidly sedimented (5 s at 12,000 ϫ g) and resuspended in 0.5 ml of 50 mM Tris-HCl (pH 7.4), 0.15 M NaCl, 1% (v/v) Igepal, 0.5% (w/v) sodium deoxycholate, and 0.1% (w/v) SDS, supplemented with proteinase inhibitors (11,15). Further treatment of the samples and immunoprecipitation with the monoclonal anti-pan cytokeratin antibody bound to protein A-Sepharose were performed as described (11,15).
Determination of the Stoichiometry of Cytokeratin Phosphorylation-The stoichiometry of cytokeratin phosphorylation was calculated by simultaneously determining (i) the specific radioactivity of the ␥-phosphate of intracellular ATP, (ii) the amount of 32 P incorporated into the cytokeratin bands, and (iii) the mass of protein in those cytokeratin bands.
(i) To determine the specific radioactivity of the ␥-phosphate of intracellular ATP, hepatocytes were labeled with 32 P for 60 min to achieve steady-state labeling of proteins before addition of the agonists (16). After the indicated times, 1.0 ml of cells was precipitated with 0.15 ml of 2 M HClO 4 . After neutralization with K 2 CO 3 , samples were centrifuged (20,000 ϫ g, 15 min). Supernatants were filtered through a filter of 0.22-m pore diameter and subsequently used for nucleotide separation exactly as described by Gualix et al. (17). The specific radioactivity of nucleotides was determined by measuring in parallel the nucleotide concentration after transformation of the A 259 peak areas to masses by correlation with standards and the 32 P incorporation into those peaks (18). The specific activity of the ␥-phosphate of ATP was taken to be the difference between the specific activities of ATP and ADP (18,19).
(ii) To determine the amount of 32 P incorporated into cytokeratins, immunoprecipitates were obtained and treated exactly as described above, the two labeled bands in the gels were cut out and their radioactivity was determined.
(iii) To determine the amount of protein in the labeled bands corresponding to cytokeratins 8 and 18, immunoprecipitates were obtained exactly as described above except that [ 32 P]P i was omitted from the hepatocyte incubation medium. Immunoprecipitates were subjected to SDS-PAGE together with varying concentrations of purified cytokeratin 8 and cytokeratin 18. Bands were visualized by the silver staining technique (20), and the amount of protein in the cytokeratin bands was calculated by interpolating the values of optical density of the immunoprecipitated protein bands in the standard curve of protein mass versus optical density constructed with the commercial, purified cytokeratins 8 and 18. The standard curve was corrected for the contaminating proteins present in the commercial preparations of purified cytokeratins 8 and 18. In order to improve the visualization of the stained protein bands of the immunoprecipitate (e.g. Fig. 6), the amount of immunoprecipitate applied for determination of cytokeratin mass was routinely 3 times the amount used for other purposes (e.g. cytokeratin phosphorylation, Western blotting). Values of cytokeratin mass inferred were corrected for this factor.
Incubation with Protein Kinases-Permeabilized hepatocytes or isolated mitochondria (1.5-2.0 mg of protein/ml), supplemented or not with cytoskeletal fractions (as indicated in every case), were incubated at 30°C for 10 min in either of the following phosphorylation media, and aliquots of the incubations were subsequently taken to determine CPT-I activity as described above. (i) Ca 2ϩ /CM-PKII phosphorylation medium contained 50 mM Hepes/KOH (pH 7.4), 100 mM KCl, 1 mM CaCl 2 , 10 mM MgCl 2 , 0.1 mM ATP, 30 ng/l calmodulin, 50 nM OA, and 0.6 ng/l purified Ca 2ϩ /CM-PKII, essentially as recommended by the supplier. (ii) cAMP-dependent protein kinase phosphorylation medium was exactly as described before (10). (iii) Protein kinase C phosphorylation medium contained 3 milliunits/l purified protein kinase C and the assay components according to the supplier.
Determination of Malonyl-CoA Concentration-Intracellular levels of malonyl-CoA were determined in neutralized perchloric acid extracts by a radioenzymatic method (15).
Statistical Analysis-Unless otherwise indicated, results shown represent the mean Ϯ S.D. of the number of hepatocyte preparations indicated in each case. Incubations of hepatocytes or mitochondria as well as enzyme assays were always carried out in triplicate. Statistical analysis was performed by Student's t test.

Effect of Mild Trypsin Digestion on CPT-I Activity-
In a first set of experiments aimed at determining whether extramitochondrial components may be involved in the control of CPT-I activity (12), permeabilized hepatocytes were treated with trypsin in very mild conditions (low doses, 4°C, 2 min) and CPT-I activity was subsequently determined. As shown in Fig.  1, when hepatocytes were incubated without further additions and then permeabilized with digitonin, trypsin was able to stimulate CPT-I by approximately 50% in these cell ghosts. Preincubation of hepatocytes with OA led to a similar activation of CPT-I in the permeabilized cell system (Fig. 1). However, trypsin was unable to produce any further stimulation of CPT-I in ghosts prepared from OA-pretreated hepatocytes (Fig.  1). The cytoskeletal stabilizer taxol has been shown to prevent the changes in hepatic CPT-I activity induced by a number of cellular effectors including OA (12). Likewise, when hepatocytes were pretreated with OA in combination with taxol, the stimulatory effect of trypsin was evident (Table I).
To test whether trypsin may cleave CPT-I itself under these digestion conditions, mitochondria were isolated from control and trypsin-treated permeabilized hepatocytes and CPT-I was subsequently detected by Western blotting (Fig. 1). As expected (5), a major band with a molecular mass of 88 kDa was detected in the blots. In addition, no differences were observed between the two preparations of mitochondria (Fig. 1). Thus, the value of relative optical density of the 88-kDa band of trypsin-treated samples was 100 Ϯ 4% (n ϭ 4), setting at 100% the value for control mitochondria. Furthermore, CPT-I activity was determined in mitochondria isolated from permeabilized hepatocytes that had been treated with or without trypsin. As shown in Table I, no differences in CPT-I activity were evident among the different conditions.
It is worth noting that treatment of permeabilized hepatocytes with trypsin had no effect on the recovery of total permeabilized cell or total mitochondrial protein. Thus, when permeabilized hepatocytes at 1.6 Ϯ 0.2 mg of protein (n ϭ 4) were treated without or with 17.5 g of trypsin for 2 min at 4°C and ghosts were collected after stopping trypsin action as described in legend to Fig. 1, 1.5 Ϯ 0.1 and 1.5 Ϯ 0.2 mg of ghost protein were recovered, respectively. Likewise, when mitochondria were isolated from those ghosts, 0.18 Ϯ 0.04 and 0.19 Ϯ 0.06 mg of protein were recovered in mitochondria prepared from trypsin-treated and trypsin-untreated ghosts, respectively.
Since permeabilized hepatocytes also express CPT activity from peroxisomes and microsomes (cf. Refs. 1-5), the contribution of CPT-I to total hepatocellular tetradecylglycidate-sensitive CPT activity was quantified. Thus, hepatocytes were incubated with 10 M tetradecylglycidate for 30 min; purified mitochondria, peroxisomes, and microsomes were isolated (9), and CPT activity was measured in these fractions. It turned out that at least 85% of total tetradecylglycidate-sensitive CPT activity experimentally determined routinely corresponded to CPT-I, whereas microsomal CPT and peroxisomal CPT together made a minor contribution (Ͻ15%) to the tetradecylglycidate-sensitive CPT pool under these conditions. Therefore, we believe that determination of CPT-I activity by our permeabilized hepatocyte procedure is not prone to substantial error.
Effect of Disruptors of the Cytoskeleton on CPT-I Activity-OA and other phosphatase inhibitors produce hyperphosphorylation and consequently disruption of the cytoskeletal network in several cell types, including hepatocytes (e.g. Refs. 21 and 22). To test whether changes in the organization of the cytoskeleton may be related to parallel changes in CPT-I activity, hepatocytes were incubated with colchicine (a microtubule   (23,24). As shown in Table II, neither colchicine nor cytochalasin B affected CPT-I activity. As a control to prove the biological activity of these two compounds in hepatocytes, cellular lipids were prelabeled with [ 14 C]palmitate, and very low-density lipoprotein output into the medium was monitored (25). As described previously (25), disruption of microtubules with colchicine or disruption of microfilaments with cytochalasin B led to a strong inhibition (Ͼ90%) of the output of very low-density lipoprotein lipids into the medium and to a parallel accumulation of intracellular lipids.
In contrast to colchicine and cytochalasin B, IDPN produced a significant increase in CPT-I activity (Table II). Interestingly, the effects of IDPN and OA were basically non-additive (Table  II). Furthermore, stabilization of the cytoskeleton with taxol prevented the stimulation of CPT-I induced by IDPN and OA, either alone or in combination (Table II). It should be pointed out that neither taxol nor IDPN changed by themselves the malonyl-CoA concentration in hepatocytes (Table II). In addition, neither of these two compounds affected the OA-induced decrease of intracellular malonyl-CoA levels ( Table II).
Effect of Cytoskeletal Fractions on CPT-I Activity-To further support the notion that cytoskeletal components may inhibit CPT-I, two cytoskeletal fractions were isolated from rat hepatocytes to study their possible inhibitory effect on CPT-I. The composition of these two fractions is shown in Fig. 2. Thus, the total cytoskeleton fraction (Fraction I) contained the major components of microtubules (tubulin), microfilaments (actin), and intermediate filaments (cytokeratins 8 and 18). The bands in the blots showed the mobilities expected for the molecular weights of the respective proteins (14). In the case of cytokeratin 8, a minor band of 45 kDa (perhaps cytokeratin 18) was sometimes detected in the blots (Fig. 2C), whereas a minor band of 54 kDa (perhaps cytokeratin 8) was sometimes detected in the blots of cytokeratin 18 (Fig. 2D).
In contrast to Fraction I, the fraction that was intended to be more enriched in intermediate filaments (Fraction II) actually contained more cytokeratins that Fraction I and was practically devoid of actin and tubulin (Fig. 2). In particular, the content of cytokeratins (relative to total protein content) of Fraction II was 4.0 Ϯ 1.3 (cytokeratin 8) and 3.7 Ϯ 0.8 (cytokeratin 18) times that of Fraction I (n ϭ 3).
Isolated rat liver mitochondria were then incubated with the cytoskeletal fractions and CPT-I activity was subsequently determined. As shown in Fig. 3, the two cytoskeletal fractions produced a dose-dependent inhibition of CPT-I activity. In agreement with the effect of IDPN described above, the fraction that was more enriched in intermediate filament components (Fraction II) produced a more potent inhibition of CPT-I (Fig.  3). Fifty percent inhibition of CPT-I by the two fractions occurred at total cytoskeletal protein:total mitochondrial protein ratios of 0.104 Ϯ 0.025 and 0.032 Ϯ 0.008 for Fraction I and Fraction II, respectively. (26), the OA-dependent stimulation of CPT-I has been suggested to rely on the phosphorylation and subsequent activation of Ca 2ϩ /CM-PKII (11). Hence, purified autophosphorylated Ca 2ϩ /CM-PKII was directly added to isolated mitochondria or permeabilized hepatocytes and CPT-I activity was determined. Ca 2ϩ /CM-PKII was able to significantly (p Ͻ 0.01) stimulate CPT-I in permeabilized cells (140 Ϯ 8% stimulation, n ϭ 4) but not in isolated mitochondria (6 Ϯ 7% stimulation, n ϭ 4). In contrast, addition of purified cAMP-dependent protein kinase or protein kinase C to permeabilized hepatocytes did not produce any change in CPT-I activity in either isolated mitochondria or permeabilized hepatocytes (data not shown).

Effect of Purified Protein Kinases on CPT-I Activity-On the basis of the antagonistic effect exerted by KN-62, an inhibitor of Ca 2ϩ /CM-PKII
In order to define the cell components that are sufficient for the malonyl-CoA-independent control of CPT-I to be demonstrated, we next attempted to reconstitute the whole-cell experimental system in a simple manner by incubating isolated mitochondria together with cytoskeletal Fraction II and purified Ca 2ϩ /CM-PKII. As shown in Fig. 4, the inhibition of CPT-I produced by exposure of isolated mitochondria to cytoskeletal Fraction II was reverted by addition of exogenous Ca 2ϩ /CM-PKII.
Phosphorylation of Cytokeratins in Intact Hepatocytes-To obtain further evidence for a possible connection between Ca 2ϩ / CM-PKII and intermediate filaments, hepatic cytokeratin phosphorylation was investigated. The phosphorylation pattern of purified cytokeratins in vitro may not reflect their phosphorylation status in more physiological, intact cell systems (27)(28)(29). Therefore, intact hepatocytes were labeled with 32 P i and cytokeratins were immunoprecipitated. As shown in Fig. 5, two major cytokeratin bands were phosphorylated upon hepatocyte challenge to OA. These two bands were assigned to cytokeratins 8 and 18 on the basis of their molecular mass (54 and 45 kDa, respectively) and high abundance in rat liver (e.g. Refs. [27][28][29]. Moreover, the OA-induced phosphorylation of these two bands was prevented by KN-62, the Ca 2ϩ /CM-PKII and malonyl-CoA levels Hepatocytes were preincubated for 45 min in the absence or presence of the modulators of the integrity of the cytoskeleton as indicated. Incubations were continued for 15 min additional with or without 0.5 M OA, and then aliquots were taken to determine the level of malonyl-CoA as well as the activity of CPT-I by the one-step assay. One-hundred percent values of CPT-I activity and malonyl-CoA concentration were 1.29 Ϯ 0.22 nmol/min ϫ mg of protein and 73 Ϯ 12 pmol/mg of protein, respectively. Results correspond to the number of experiments indicated in parentheses for CPT-I activity and to three different experiments for malonyl-CoA concentration.  inhibitor that antagonizes the OA-induced stimulation of CPT-I (11). Nevertheless, the Ca 2ϩ ionophore A23187 had no effect on cytokeratin phosphorylation in intact hepatocytes (Fig. 5, lanes e and f).
To confirm that the 32 P-labeled bands do, indeed, correspond to cytokeratins 8 and 18, and not to proteins which have coprecipitated with the cytokeratins, a Western blotting analysis of the immunoprecipitated proteins was performed with the anti-cytokeratin 8 and anti-cytokeratin 18 monoclonal antibodies. The rationale of this experiment was that it would be most unlikely that proteins different to cytokeratins would also cross-react with the anti-cytokeratin antibodies on a Western blot, especially when, as in the present study, different sources of antibodies are used in the immunoprecipitation and in the blotting. As shown in Fig. 6, the 54-kDa band actually contained cytokeratin 8, whereas cytokeratin 18 was actually present in the 45-kDa band.
Stoichiometry of Cytokeratin Phosphorylation-Since cytokeratins are rather abundant elements of the hepatocyte cytoskeleton (27)(28)(29), it might be argued that the stoichiometry of cytokeratin phosphorylation might be very low and therefore non-functional. Therefore, the phosphorylation state of cytokeratins in okadaic acid-treated hepatocytes was determined. For this purpose, we determined the specific radioactivity of the ␥-phosphate of intracellular ATP by a high performance liquid chromatography method, the amount of 32 P incorporated into the cytokeratins, and the amount of cytokeratins in the immunoprecipitated bands as described under "Experimental Procedures." This calculation assumes that all of the phosphate in the phosphorylated proteins reaches isotopic equilibrium with the ␥-phosphate of intracellular ATP after the 60-min labeling period with [ 32 P]P i (16,18). After this 60-min labeling period, the ratio of specific radioactivities of ATP:ADP:AMP in two separate experiments was 1.00:0.62:0.15 and 1.00:0.56: 0.11 (Table III). These values are in agreement with those obtained by Holland et al. (18) and indicate that the ␤and the ␥-phosphates are at isotopic equilibrium with each other, but not with the ␣-phosphate. Fig. 6 shows a gel used for the calculation of the amount of protein in the 54-and 45-kDa bands. Values of mole of phosphate/mole of cytokeratin obtained in two separate experiments (Table III) are in agreement with the observations that cytokeratins 8 and 18 (and mostly the former) become significantly phosphorylated upon exposure of intact hepatocytes to phosphatase inhibitors (22,28), although in the latter two reports no quantification of cytokeratin phosphorylation was achieved. Therefore, cytokeratin phosphorylation may be functionally relevant in our system. tion suggest that CPT-I may become activated by cleavage of extramitochondrial (although not necessarily cytoskeletal) cell component(s). In line with this observation, Fontaine et al. (32) have recently reported that porin, the mitochondrial outermembrane pore-forming protein, also becomes activated in permeabilized hepatocytes upon mild trypsin digestion of extramitochondrial cell components. It is worth noting that the digestion conditions employed in the present paper were much milder than those previously used by Kashfi and Cook (33) to study the effect of proteolysis on CPT-I, and, therefore, the two types of experiments are not comparable. In line with our observations, Fraser et al. (34) did not observe any effect of trypsin on CPT-I under digestion conditions more or less comparable to ours. Interestingly, cell pretreatment with OA prevented further activation of CPT-I by trypsin, suggesting that both OA and trypsin may share a common mechanism to relieve CPT-I from inhibition. Second, incubation of intact hepatocytes with IDPN increased CPT-I activity in a basically nonadditive manner with respect to OA, suggesting a common mechanism of action. Third, CPT-I activity in isolated mito-chondria was depressed in a dose-dependent fashion by the addition of a total cytoskeleton fraction and a cytokeratinenriched cytoskeletal fraction, the latter being 3 times more potent than the former. Fourth, taxol prevented the OA-induced desensitization of CPT-I to trypsin activation, as well as the OA-and IDPN-induced stimulation of CPT-I. In short, all these data suggest that disruption of interactions between CPT-I and cytoskeletal component(s) may de-inhibit CPT-I and, therefore, increase enzyme activity.

Involvement of Cytoskeletal Components in the Control of CPT-I Activity-
The possibility that CPT-I interacts with cytoskeletal components as put forward in this paper is in line with the current notion that the dynamics of mitochondria in living cells may result from specific interactions of mitochondria with components of the cytoskeleton (30,31,(35)(36)(37)(38). It has been suggested that a function of the interactions between mitochondria and intermediate filaments may be to locate mitochondria in precise sites within the cell (30,31,39). This idea is based on both in vitro (39,40) and in vivo (41,42) experiments. Since the organization of intermediate filaments changes dramatically in a number of liver pathologies (43), the observations described in the present paper predict that CPT-I activity as affected by cytoskeletal components may change under pathophysiological situations in which the organization of the cytoskeleton is altered, e.g. in transformed cells. In this respect, Paumen et al. (44) have recently observed that inhibition of CPT-I with etomoxir leads to a stimulation of ceramide synthesis and to palmitate-induced cell death. These authors suggested that cells that express high CPT-I activity are expected to withstand palmitate-induced apoptosis (44). Thus, we have recently observed that CPT-I specific activity is similar in mitochondria isolated from hepatoma cells and normal hepatocytes, but just about half in permeabilized hepatocytes than in permeabilized hepatoma cells; in addition, CPT-I is not activated by OA in hepatoma cells (45). These observations support the notion that in hepatocytes OA liberates CPT-I from certain constrictions imposed by extramitochondrial cell components that do not operate either in isolated mitochondria or in transformed liver cells. Whether liberation of CPT-I from those potential constrictions may help hepatoma cells to escape from apoptosis is currently under study in our laboratories. Anyway, it is worth noting that treatment of hepatocytes with OA, a well known tumor promoter, renders a "CPT-I regulatory phenotype" similar to that shown by hepatoma cells.
Involvement of Ca 2ϩ /CM-PKII in the Control of CPT-I Activity-Previous experiments in our laboratories have shown that KN-62, an inhibitor of Ca 2ϩ /CM-PKII (26), antagonizes the OA-induced stimulation of hepatic CPT-I activity (11). Data in the present report further point to a link between Ca 2ϩ /CM-PKII and the cytoskeleton in the context of CPT-I regulation. This conclusion is based mostly on three observations. First, purified Ca 2ϩ /CM-PKII was able to activate CPT-I in permeabilized cells but not in isolated mitochondria. This is in agreement with previous evidence against the involvement of direct phosphorylation in the OA-induced stimulation of CPT-I (10) and indicates that extramitochondrial cell components are required for the regulation of CPT-I activity by Ca 2ϩ /CM-PKII. In this respect, it is worth noting that permeabilization of hepatocytes with digitonin seems to preserve quite well both the general morphology of the cell and the structure of the cytoskeleton (46), and, therefore, the potential interactions between the cytoskeleton and cell organelles may remain basically unaffected upon this type of manipulation. Second, when isolated mitochondria were incubated with a cytokeratinenriched cytoskeletal fraction, purified Ca 2ϩ /CM-PKII was able to abrogate the inhibition of CPT-I induced by that cytokeratin fraction. It is clear from these experiments that a

Stoichiometry of cytokeratin phosphorylation in isolated hepatocytes
Hepatocytes were labeled by incubation for 1 h with 32 P and further incubated for 15 min in the presence of 0.5 M OA. Aliquots of the incubations were taken to determine the specific radioactivity of adenine nucleotides and the radioactivity incorporated into cytokeratin 8 (CK8) and cytokeratin 18 (CK18). Parallel incubations were performed in 32 P-free medium to determine the amount of protein in the bands corresponding to CK8 and CK18 (see Fig. 6 simple reconstituted system composed of isolated mitochondria, a cytokeratin-enriched fraction, and purified Ca 2ϩ /CM-PKII may reflect the situation occurring in the intact hepatocyte, indicating that these three components are sufficient for the malonyl-CoA-independent acute control of CPT-I to be demonstrated in vitro. Third, the Ca 2ϩ /CM-PKII inhibitor KN-62 prevented the OA-induced phosphorylation of cytokeratins in intact hepatocytes, pointing to a role of Ca 2ϩ /CM-PKII on cytokeratin phosphorylation in these cells (28,29). Additional evidence for the involvement of cytokeratins in the control of CPT-I activity is given by the lack of effect of A23187 on cytokeratin phosphorylation in hepatocytes. In this context, challenge of hepatocytes to OA leads to CPT-I activation and cytokeratin phosphorylation, whereas elevation of cytosolic free Ca 2ϩ concentration by A23187 has no effect on either CPT-I activity (47) or cytokeratin phosphorylation (the present paper). The possibility that liver Ca 2ϩ /CM-PKII has a different pattern of activation by Ca 2ϩ /calmodulin and by autophosphorylation than brain Ca 2ϩ /CM-PKII (cf. Ref. 11) is as yet an open question.
It is worth noting that neither cAMP-dependent protein kinase nor protein kinase C affected CPT-I activity in permeabilized hepatocytes. This is in line with the observation that neither cAMP-dependent protein kinase inhibitors nor protein kinase C inhibitors were able to prevent the OA-induced stimulation of CPT-I (11). As a matter of fact, several reports indicate that, despite their ability to phosphorylate cytokeratins in vitro, neither of these two protein kinases play an important role in the direct control of intermediate filament integrity in intact hepatocytes (28,48,49). In contrast, and in line with data in the present paper, Ca 2ϩ /CM-PKII has been shown to play a major role in the phosphorylation and functional integrity of hepatic cytokeratins in vivo (28) as well as in the OA-induced disruption of hepatic cytoskeleton (21).
Malonyl-CoA-dependent and Malonyl-CoA-independent Control of CPT-I Activity-Together with previous observations (10 -12), data in this paper allow for a model that explains the OA-induced malonyl-CoA-independent control of hepatic CPT-I. As shown in Scheme I, OA may activate Ca 2ϩ /CM-PKII by increasing its degree of phosphorylation upon inhibition of protein phosphatases 1 and 2A; this effect would be overcome by KN-62, an inhibitor of Ca 2ϩ /CM-PKII autophosphorylation. Activated Ca 2ϩ /CM-PKII would phosphorylate cytoskeletal components, perhaps cytokeratins 8 and 18, thereby disrupting putative inhibitory interactions between the cytoskeleton and CPT-I. Stimulation of CPT-I upon disruption of the cytoskele-ton would be also achieved by challenge of intact hepatocytes to IDPN or by treatment of permeabilized hepatocytes with trypsin in mild conditions. Stabilization of the cytoskeleton with taxol may prevent the malonyl-CoA-independent acute stimulation of CPT-I.
It is obvious that the notion that fatty acid translocation into mitochondria may be controlled by modulation of the interactions between CPT-I and cytoskeletal components (i.e. by a malonyl-CoA-independent mechanism) does not diminish the importance of malonyl-CoA as a physiological modulator of CPT-I activity (5,8). On the one hand, since the pioneering work of McGarry and co-workers (8,50), changes in long-chain fatty acid oxidation under many different pathophysiological situations have been shown to be linked to changes in intracellular malonyl-CoA concentration and/or changes in the sensitivity of CPT-I to malonyl-CoA (1,2,5). On the other hand, several observations suggest that malonyl-CoA-dependent and malonyl-CoA-independent acute control of hepatic CPT-I activity might operate in concert. First, we have recently shown that stimulation of the AMP-activated protein kinase, a major protein kinase involved in the control of hepatic lipid metabolism, leads to an activation of hepatic CPT-I by malonyl-CoA-dependent and malonyl-CoA-independent mechanisms (51). Second, a fraction of hepatic acetyl-CoA carboxylase, the enzyme responsible for the synthesis of malonyl-CoA, has been recently suggested to be bound to the cytoskeleton (52). Third, it has been put forward that the 280-kDa isoform of acetyl-CoA carboxylase might interact with the outer leaflet of the mitochondrial outer membrane in order to channel malonyl-CoA for CPT-I inhibition (53). Fourth, the recent observation that the bulk of the CPT-I protein seems to face the cytoplasmic side of the mitochondrial outer membrane (34) makes more likely that interactions between CPT-I and cytoskeletal components might occur. Although the physiological role of the malonyl-CoA-independent mechanism of regulation of hepatic CPT-I activity is as yet unknown, it is worth noting that hormonal challenge of hepatocytes (e.g. glucagon, insulin) leads to changes in CPT-I activity that parallel changes in long-chain fatty acid oxidation and that are retained after washing of the permeabilized cells (3). In the context of the emerging role of cytoskeletal filamentous networks in intracellular signaling (54), current research in our laboratories is focussed on the possible existence of a coordinate control of CPT-I and acetyl-CoA carboxylase activities by modulation of interactions between the cytoskeleton and the mitochondrial outer membrane.