Dietary and biliary phosphatidylcholine activates PKCζ in rat intestine.

Chylomicron output by the intestine is proportional to intestinal phosphatidylcholine (PC) delivery. Using five different variations of PC delivery to the intestine, we found that lyso-phosphatidylcholine (lyso-PC), the absorbed form of PC, concentrations in the cytosol (0 to 0.45 nM) were proportional to the input rate. The activity of protein kinase C (PKC)ζ, which controls prechylomicron output rate by the endoplasmic reticulum (ER), correlated with the lyso-PC concentration suggesting that it may be a PKCζ activator. Using recombinant PKCζ, the Km for lyso-PC activation was 1.49 nM and the Vmax 1.12 nM, more than the maximal lyso-PC concentration in cytosol, 0.45 nM. Among the phospholipids and their lyso derivatives, lyso-PC was the most potent activator of PKCζ and the only one whose cytosolic concentration suggested that it could be a physiological activator because other phospholipid concentrations were negligible. PKCζ was on the surface of the dietary fatty acid transport vesicle, the caveolin-1-containing endocytic vesicle. Once activated, PKCζ, eluted off the vesicle. A conformational change in PKCζ on activation was suggested by limited proteolysis. We conclude that PKCζ on activation changes its conformation resulting in elution from its vesicle. The downstream effect of dietary PC is to activate PKCζ, resulting in greater chylomicron output by the ER.

groups (A to E) of rats participated in the experiments. All the animals were maintained on a chow diet (4% fat, w/w) except group D, which was placed on a high-fat diet (23% fat, w/w) for 2 weeks prior to the experiment. All rats received duodenal cannulas (PE 50) 1 day prior to the experiments and were infused intraduodenally with 0.15 M NaCl, 0.3 M KC1, and 5% glucose at 3 ml/h overnight. Group A rats had their bile duct cannulated (PE 10) in addition to their duodenal cannulas. The next day, the infusion for groups A and B was changed to 0.15 M NaCl. The infusions for groups C and D were changed to 30 mM triolein (TO), 10 mM taurocholate (TC), 0.15 M NaCl, and 10 mM Tris-HC1 (pH 7.4). Group E was given the same TO infusion supplemented with 2 mM PC. All infusions were at 4.5 ml/h for 4 h and were supplemented with [ 3 H]oleate TO (10 × 10 6 dpm/ml). A summary of the models used is presented in Table 1 . After the 4 h infusion was completed, all of the groups were treated identically. The rats were given an overdose of pentobarbital, and the proximal one-half of the small intestine was removed, fl ushed with ice-cold 0.15 M NaC1, and placed in iced saline.

Isolation of cytosol
Enterocytes from the proximal half of rat small intestine were isolated as described previously ( 1 ). In brief, enterocytes were isolated from intestinal villi ( 15 ) homogenized using a Parr bomb, and the cytosol was isolated by centrifugation. The isolated cytosol was dialyzed against buffer A [0.25 M sucrose, 30 mM HEPES (pH 7.2), 30 mM KCl, 5 mM MgCl 2 , 5 mM CaCl 2 , 2 mM DTT] overnight at 4°C. This cytosol was concentrated on a Centricon fi lter (Amicon, Beverly, MA) with a 10 kDa cutoff to 20 mg of protein/ml.

Isolation of caveolin-1-containing endocytic vesicles from cytosol
Caveolin-1-containing endocytic vesicles (CEVs) were isolated as described previously ( 16 ). In brief, cytosol (1 mg protein) was treated with 1% Triton X-100. The treated cytosol was placed at the bottom of an OptiPrep step gradient (0 to 35%), and the gradient was centrifuged for 20 h at 4°C. The gradient was resolved by collecting nine 1 ml fractions using a pipette starting at the top of the gradient. One hundred microliters from each fraction was collected for radioactivity determination. The proteins in each fraction were precipitated with TCA, washed twice with cold acetone, and suspended in Laemmli's buffer. The proteins were separated by SDS-PAGE and identifi ed by immunoblot using specifi c antibodies as indicated.

H]PC in CEV after corn oil gavage
Nonfasting rats were anesthetized and gavaged with 0.5 ml corn oil containing [ 3 H]PC (12 × 10 6 dpm). Thirty minutes later, the rats were euthanized and the proximal half intestines removed. Enterocytes were isolated, and cytosol was obtained as described above. CEVs were isolated by OptiPrep gradient. Ninety-two percent of the cytosolic radioactivity was in lyso-PC.

Extraction of phospholipids from cytosol
Phospholipids were extracted from cytosol by chloroformmethanol extraction. Two milliliters of CHCl 3 /methanol (2:1) was added to 100 l of cytosol and agitated vigorously. The extracts were centrifuged at 9000 g for 5 min at 4°C. Five hundred microliters of 0.9% NaCl was added to the bottom layer, mixed vigorously, and recentrifuged as described above. The chloroform layer was collected and evaporated under nitrogen for quantitation. PKC typically contain phosphatidylserine (PS) as activator. Because the amount of PS required for PKC activation is greater than that present in intestinal cytosol, we hypothesized that a different activator was operative. The hypothesis tested in this report is that lyso-phosphatidylcholine (lyso-PC) is the physiological activator of PKC in the intestine. We focused on lyso-PC because it has long been known that the amount of phosphatidylcholine (PC) delivered to the intestine correlates with chylomicron output into the lymph ( 9 ). Prior to its absorption, PC must be hydrolyzed to sn -1 lyso-PC by pancreatic phospholipase A 2 . In accord with this hypothesis, bile diversion or mdr2 knockout mice, which are not able to deliver PC to the bile, are associated with the lowest chylomicron output by the intestine ( 10, 11 ); whereas fat feeding, which increases biliary PC output ( 12 ), and PC supplementation both increase chylomicron output ( 13 ) as compared with chowfed controls. These data were obtained using the same intraduodenal TAG input rates. In the past, the correlation between chylomicron output and PC delivery to the intestine has been thought to be due to the availability of PC for chylomicron surface formation. We provide evidence of an alternative mechanism in this report.

Antibodies
Rabbit polyclonal antibodies against CD36, caveolin-1, PKC , and goat polyclonal antibodies against intestinal alkaline phosphatase were all purchased from Santa Cruz Biotechnology (Santa Cruz, CA). PKC substrate-biotinylated was obtained from Santa Cruz Biotechnology. Biotinylated Anti-Rabbit IgG (H+L) antibody was purchased from KPL (Gaithersburg, MD). Streptavidin, HRP conjugate was procured from EMD Millipore Co. (Billerica, MA). Anti-rabbit IgG conjugated with agarose beads was purchased from Sigma Chemical Co. Goat anti-rabbit IgG and goat anti-mouse IgG conjugated with HRP were also procured from Sigma. PKC was purchased from Enzo Life Sciences Inc. (Farmingdale, NY).

Animal preparation
Intraduodenal infusion of male Sprague-Dawley rats (200-300 g) was performed as described previously ( 14 ). In brief, fi ve 5 mM EDTA, 10 mM EGTA, 50 mM ␤ -mercaptoethanol), 15 l of 1 mM ATP, and 65 l water for 30 min at 4°C. The reaction was stopped by adding 500 l of cold PBS. CEVs were reisolated by OptiPrep step gradient as described above. Repopulation of CEVs by activated PKC was performed by the incubation of 500 g PKC depleted cytosol with 1 nM of activated rPKC (incubated with 6 nM lyso-PC) in a total volume of 500 l buffer D (0.25 M sucrose, 30 mM HEPES, pH 7.2, 30 mM KCl, 5 mM MgCl 2 , 5 mM CaCl 2 , 2 mM DTT ) with an ATP regenerating system (1 mM ATP, 5 mM phosphocreatine, and 5 U creatine phosphokinase) for 30 min at 4°C. CEVs were reisolated using an OptiPrep gradient. The proteins from each fraction were separated by SDS-PAGE and identifi ed by immunoblot using PKC antibodies.

Proteolysis method
Native and lyso-PC-activated PKC were subjected to tryptic digestion. Digestions were carried out at 37°C for 0, 2, 4, 8 min by using an enzyme-protein mass ratio of 1:200. Reactions were terminated by adding Laemmle's buffer and boiled for 5 min. The proteolytic digestion products were resolved on SDS-PAGE followed by silver staining.

Immunoprecipitation of proteins
Cytosol (1 mg) was incubated with 20 l of specifi c rabbit antibodies at 4°C for 4 h, anti-rabbit IgG conjugated with agarose beads was added, and the mixture was stirred slowly at 4°C overnight. The beads were removed by centrifugation, washed 10 times with cold PBS, and resuspended in Laemmli's buffer for immunoblot analysis.

SDS-PAGE and immunoblot
Proteins were separated by SDS-PAGE and then transblotted to nitrocellulose membranes (Bio-Rad). After incubation with specifi c primary antibodies and peroxidase-conjugated secondary antibodies, labeled proteins were detected using ECL (Amersham Biosciences) and Biomax fi lm (Eastman Kodak, Rochester, NY).

Measurement of TAG and phospholipid radioactivity
TAG radioactivity was determined by liquid scintillation as described ( 16 ). Phospholipids were similarly measured.

[ 3 H]oleate is absorbed via CEV under varying PC infusion conditions
Because we planned to greatly vary the PC input rate into the intestine, we fi rst wished to be certain that the variation either in the delivery of PC to the intestinal lumen in our models or the fl ux of lyso-PC into the intestinal

Quantitation of lyso-PC
Phospholipids in cytosol were fi rst quantifi ed as described by McHowat and Corr ( 17 ). In brief, samples were injected onto an Ascentis Si 5-m HPLC (25 cm × 4.6 mm) column (Supelco Analytical, Sigma Chemical Co.) and eluted with a mobile phase of hexane-isopropyl alcohol-water (45:50:5) at a fl ow rate of 1.0 ml/ min. Absorbance was measured at 203 nm. Samples were also quantitated radiometrically using [ 14 C]acidic anhydride (10 mCi/ mM; ARC, St. Louis, MO). Samples were treated as suggested by Wientzek et al. ( 18 ). In brief, the lipids from cytosol were extracted as above, and the lyso-PC was isolated from a silicic acid column using chloroform-methanol (1:9) after other lipids had been removed by passing progressively more methanol in chloroform over the column. The lyso-PC was treated with [ 14 C]acidic anhydride (20,000 dpm), and the acetylated lyso-PC was separated by TLC. Its radioactivity was determined and compared with known quantities of lyso-PC treated similarly.

PKC activity assays
PKC activity in cytosol was quantifi ed using a modifi ed method described by Calbiochem (EMD Millipore Co.). In brief, to bind biotinylated anti-rabbit antibody to agarose beads, we incubated 10 l biotinylated anti-rabbit antibody with 100 l of anti-rabbit agarose beads for 3 h at 4°C. After incubation, the beads were washed three times with PBS. Recombinant PKC (rPKC ) in 100 l of buffer B (250 mM Tris-HCl, pH 7.0, 30 mM MgCl 2 , 5 mM EDTA, 10 mM EGTA, 50 mM ␤ -mercaptoethanol), 15 l of 1 mM ATP, 15 l of 0 to 40 nM PS or 0 to 10 nM of lyso-PC, and 65 l water were incubated with 2 l of biotinylated PKC substrate for 5 min at 37°C. The reaction was stopped with 500 l of cold PBS. The reaction mixture was added to anti-rabbit agarose beads as described above for 3 h at 4°C. After incubation, the beads were washed with cold PBS three times and suspended in 400 l of PBS. Five microliters of HRP-conjugated streptavidin was added to the suspension and incubated for 2 h at 4°C, washed three times with cold PBS, followed by the addition of 100 l of substrate (5 mg o -phenylenediamine in 5 ml of H 2 O 2 ), and incubated for 5 min at room temperature. The reaction was stopped with 100 l of 20% H 3 PO 4 , and 2 ml of PBS was added. The supernatant was obtained by centrifugation and read at 492 nm.

Immunodepletion of PKC
Immunodepletion was performed by immunoadsorption ( 1 ). Briefl y, cytosol (1 mg) was incubated with 20 l of anti-PKC rabbit polyclonal antibodies at 4°C for 4 h, and then anti-rabbit IgG conjugated with agarose beads was added. The mixture was slowly agitated for 4 h at 4°C. The antibody-protein complexes were removed by centrifugation. Successful depletion of PKC from cytosol was obtained by three rounds of immunodepletion and confi rmed by immunoblotting.  distribution in the OptiPrep gradient. These data are consistent with the lyso-PC being in the CEV ( Fig. 2 ).

Lyso-PC concentrations in cytosol vary with PC delivery conditions
It would be predicted that altering the intake of lyso-PC from the intestinal lumen would be refl ected in the concentration of lyso-PC in the cytosol. To see if this were so, we measured the lyso-PC concentration in intestinal cytosol in each of our models by HPLC at 203 nm ( Fig. 3A ). To determine whether the signal at 203 nm was due solely to lyso-PC, we performed LC/MS/MS on a sample (supplementary Fig. 1). The spectrum obtained confi rmed that the signal at 203 nm was lyso-PC. Because absorption at 203 nm may in part be due to artifacts, we also performed a radiometric assay on each sample. Lyso-PC was isolated by silicic acid column chromatography and acetylated using [ 14 C]acetic anhydride. The acetylated lyso-PC was separated by TLC, and its radioactivity was determined. Similarly treated lyso-PC standards were used for cytosol did not affect the mechanism of FA absorption. Absorbed FA enters the cytosol on a CEV and appears in the detergent resistant membrane (DRM) fraction of an Op-tiPrep gradient ( 16 ). Despite the wide variation in the PC intake (0-9 µmol/h), the FA tracer, [ 3 H]oleate, remained in the DRM in cytosol, consistent with our prior observations ( Fig. 1A -E , Table 1 ). Under all experimental conditions, caveolin-1, CD36, and PKC tracked with the [ 3 H] oleate in the DRM. We have previously shown that the detergent soluble fraction in the OptiPrep gradient contains the expected clathrin ( 16 ).
These data suggest the possibility that lyso-PC would also be associated with CEV. To test this, rats were gavaged with corn oil supplemented with PC and [ 3 H]PC. After 30 min, the intestine was harvested, and the cytosolic fraction was obtained. Ninety-two percent of the total cytosolic dpm were in lyso-PC. To determine the percentage of the lyso-PC dpm in CEV, we placed the cytosol on an OptiPrep gradient. We found that the majority (69%) of the lyso-PC in cytosol was in the DRM fraction as suggested by its phosphatidylinositol, but only negligible amounts were found (data not shown). Similarly, the lyso derivatives of each phospholipid except PC were present in very small amounts (data not shown).

PKC activity varies with cytosolic lyso-PC concentration
Our major interest was whether the changes in cytosolic lyso-PC were associated with alterations in PKC activity. To determine PKC activity in our models, we fi rst wished to know the range of cytosolic protein concentrations associated with rates of PKC activity proportional to the PKC added ( Fig. 4A ). All subsequent PKC activity measurements used protein concentrations of 40 µg or less. As shown in Fig. 4B , PKC activity closely tracked the concentration of lyso-PC in the cytosol ( Fig. 3 ). The lowest activity was found in the bile diversion model, and the highest activity was found when PC was included in the intraduodenal TO infusion model with a 34-fold difference between them. These changes in enzyme activity were not due to alterations in the quantity of PKC as judged by immunoblot ( Fig. 4C ).

Lyso-PC activates PKC
The correlation of lyso-PC concentrations in the cytosol with PKC activity suggested that lyso-PC might be an activator of PKC . To directly test this, we incubated rPKC with ATP and increasing amounts of lyso-PC. The results ( Fig. 5A ) show a robust response in PKC activity on the addition of lyso-PC suggesting that lyso-PC is a potent PKC activator. Importantly, the calculated K m of 1.49 nM for lyso-PC activation is above the values found for lyso-PC concentrations in the cytosol in our models. This implies that changes in lyso-PC concentrations in our models would have a direct effect on PKC activity confi rming the data in Fig. 4 . Because PS is a well-known activator of PKC , we also tested PKC activity with increasing amounts of PS.
As expected, the lyso-PC concentration in the cytosol varied with the input rate ( Fig. 3A, B ). Because most of the PC delivered to the intestine comes from bile ( 19 ), bile diversion should result in a reduction in cytosolic lyso-PC over bile intact rats. In agreement with this expectation, only negligible amounts of lyso-PC were found in this model ( Fig. 3A, B ). By contrast, the maximal amount of lyso-PC was found when PC was included in the intraduodenal TO infusion, 0.46 nM ( Fig. 3A, B ). We also analyzed the cytosol for PC, PS, phosphatidylethanolamine, and 3 H]PC. Thirty minutes later, the proximal half of the intestine was removed, and cytosol was prepared (Materials and Methods). The cytosol was placed on an OptiPrep gradient, the gradient was resolved, and fraction was collected from the top of the gradient. Radioactivity was determined in each fraction (ordinate). Ninety-two percent of the total radioactivity in cytosol was in lyso-PC. Fig. 3. The concentration of lyso-PC in cytosol under varying input rates of PC into the intestine. Intestinal cytosol was prepared from the proximal half of the intestine using experimental conditions as in Fig. 1 , and phospholipids were extracted (Materials and Methods). A: Lyso-PC was quantitated by HPLC at 203 nm. B: Lyso-PC was separated using a silicic acid column and quantitated radiometrically after acetylation using [ 14 C]acidic anhydride (Materials and Methods). The P values above the bars test the differences between the means of the PC-infused rats with the other groups. However, each group was signifi cantly different ( P < 0.05) from the next group with more lyso-PC in the cytoplasm. As expected, PS also activated PKC ( Fig. 5B ) but with a reduced potency as compared with lyso-PC with a K m of 41 nM, 28-fold greater than lyso-PC. Importantly, the concentration of PS in our models was so low that it would not be expected to be an effective activator of PKC (data not shown).
Because lyso-PC is a detergent, we wondered if this property activated PKC or if there were a more specifi c interaction between the two. To test this, we used both SDS and Triton X-100 (1%), a concentration similar to what was used to isolate CEV. Both detergents were ineffective as activators (0.03% and 0.09% the activity of lyso-PC stimulated PKC , respectively). These results were not due to denaturation of PKC , because the enzyme maintains its ability to be activated by lyso-PC even after exposure to 1% Triton X-100. Finally, we considered the possibility that lyso-PC bound to PKC nonspecifi cally. To determine whether this were so, we immunoprecipitated PKC , Sar1b, vesicle associated membrane protein (VAMP)7, and IgG from intestinal cytosol using cytosol isolated from rats gavaged with corn oil, PC, and [ 3 H]PC. At the end of three rounds of immunodepletion, when anti-PKC antibody was used, 75% of the [ 3 H]PC had been removed from the cytosol but only 13% of the dpm when Sar1b was immunodepleted, 12% using anti-VAMP7 antibodies, and 9% using IgG. In sum, these data support the thesis that lyso-PC specifi cally binds to PKC and activates it.
To more directly confirm the relationship between PKC activation and lyso-PC concentrations in cytosol, we related PKC activation to lyso-PC concentration. This relationship revealed a typical saturation curve ( Fig. 5C ). We utilized the initial rate portion of the curve to delineate a mathematical relationship between the two variables ( Fig.  5D ). This would suggest that for every additional 1 nM of lyso-PC in the cytosol, PKC is ‫ف‬ 20% activated.

Lyso-PC is more effective as a PKC activator than other phospholipids or their lyso derivatives
When lyso-PC was directly tested against PS and ceramide as an activator of PKC , lyso-PC was more effective than either phospholipid even though 7-fold more PS and 16-fold more ceramide than lyso-PC were used in the assay ( Fig. 6A ). Importantly, FA (oleate) was also ineffective as an activator of PKC . We also tested lyso-PC against the lyso derivatives of other phospholipids. Although each was an effective activator of PKC , lyso-PC was the most potent despite the fact that the other lyso derivatives were used at least at 20-fold greater concentrations ( Fig. 6B ). Because in vivo the concentrations of the lyso compounds except for lyso-PC were very low in cytosol, none of these would be expected to be effective in vivo PKC stimulators. Further, we found that the concentration of ceramide in cytosol did not vary under changing PC input rates into the intestine ( Fig. 6C ).

PKC elutes from CEV on activation
We next wished to see what effect exposure of PKC to lyso-PC had on its binding to CEV. The data presented in   7A suggest that the greater the concentration of lyso-PC in cytosol, the more PKC eluted from CEV. The data are consistent with the hypothesis that activation of PKC detaches PKC from CEV allowing it to diffuse in the cytosol. By contrast, other components of CEV such as caveolin-1 ( Fig. 7B ) and CD36 ( Fig. 7C ) were relatively unaffected by changes in lyso-PC concentrations and remained with the CEV. These data confi rm the specifi city of the effect of lyso-PC on PKC with respect to CEV binding.
To further confi rm the effect of PKC activation on PKC binding to CEV, we tested whether PKC exposed to lyso-PC would bind to CEV. As shown in Fig. 7D , native CEV is replete with PKC (bar 1) as expected under native conditions, but on exposure to lyso-PC, CEV was nearly completely depleted of PKC (bar 2). Importantly, rPKC , activated by lyso-PC, does not repopulate CEV (bar 3). In sum, these data would suggest that activated PKC elutes from CEV in a vectorial fashion.

PKC activation is associated with a conformational change in the enzyme
The elution of PKC from CEV on its activation suggested the possibility that activation induces a conformational change in the kinase. To test this idea, we used limited proteolysis using trypsin on native and lyso-PC-activated rPKC ( Fig. 8A ). PKC was incubated with trypsin from 0 to 8 min as shown below the gel for both the native and activated species. Native PKC resisted proteolysis significantly more than activated PKC . Native PKC 's silver stain signal was extinguished at 8 min of incubation as compared with 4 min for activated PKC . Proteolytic fragments can be seen at lower Mr than PKC at each time of incubation except for T = 0. Fig. 8B, C show a graphical presentation of the densitometry readings (Gel DocX) at each incubation time as a percentage of the density reading at T = 0 for nonactivated PKC ( Fig. 8B ) and activated PKC ( Fig. 8C ). The increase in the rapidity of proteolysis on activation of PKC is clearly evident.

CD36, caveolin-1, and PKC interact in vivo
Although CD36, PKC , and caveolin-1 were each present on CEV, we next wished to determine whether they were interactive in vivo. To test this, we performed a series of immunoprecipitation experiments ( Fig. 9 ) using cytosol from PC-infused rats (see Materials and Methods). Each of the proteins studied, CD36, PKC , and caveolin-1, was found to immunoprecipitate the other two. When the reverse immunoprecipitations were performed, an equally strong signal was obtained confi rming the specifi city of the reaction.  6. PKC activity using varying potential activators. A: PKC activity using lyso-PC, PS, oleate (FA), or ceramide as activators; 6 nM, 40 nM, 1 µM, and 100 nM potential activator was added, respectively. rPKC (9 pM) was used as the enzyme source, and PKC activity was expressed as arbitrary units on the ordinate. B: PKC activity in response to various lyso-phospholipids as indicated on the abscissa. Lyso-PC (6 nM), lyso-PI (100 nM), lyso-PS (100 nM), and lyso-PE (100 nM) were used as activators. The data are expressed in arbitrary units on the ordinate. Signifi cant differences between the means are indicated by P values above the bars. C: Immunoblot using anti-ceramide antibodies using 30 µg cytosolic protein in each lane for each model (as in Fig. 1 , see Materials and Methods).

PKC binds to Sar1b
Finally, we wished to know whether PKC , once eluted from CEV, could bind to its cytosolic substrate, Sar1b. This interaction was confi rmed by the co-immunoprecipitation experiments shown in Fig. 9 . For these experiments, cytosol isolated from PC-infused rats (see Materials and Methods) was used to ensure that most PKC was eluted off CEV and was available for potential binding to Sar1b. As shown, the strongest binding to Sar1b was with PKC .

DISCUSSION
The present studies were undertaken to test the hypothesis that the physiological activator of PKC in the intestine is lyso-PC. This thesis was based on prior work that showed that chylomicron output into the lymph by the intestine was directly proportional to the amount of PC delivered to the intestine ( 10,11,13 ). Because the ability of the intestine to export chylomicrons presumably depends on the rate of budding of the PCTV, we thought it likely that lyso-PC, the absorbed hydrolytic product of dietary and biliary PC, would be an activator of PKC . This supposition was supported by our preliminary fi ndings on ER-PCTV budding activity using bile-diverted and PCinfused rats ( 20 ), and the fact that in the absence of PKC , no ER-PCTV budding occurred ( 7,8 ). Additionally, we felt that if we could correlate cytosolic lyso-PC concentrations with PCTV budding, it would more closely tie PCTV budding to chylomicron export by the intestine.
We have used primary isolates of intestinal absorptive cells for these studies. Although intestinal cell models such as the Caco2 cell and the IPEC cell have been used to study lipid absorption and transport, none of them are as effi cient as intestinal cells in vivo in the transport of dietary fat. The cell models transport <10% of their TAG content, whereas rat intestine is able to turn over 60% of its TAG content per hour ( 13 ). There are a number of potential explanations for this. Caco2 cells have greatly reduced monoacylglycerol acyltransferase (MGAT) activity ( 21 ). Acylation of MAG by MGAT provides the major source of DAG substrate for TAG synthesis. TAG synthesized from MAG is more likely to be incorporated into chylomicron-TAG than if it is synthesized via the Kennedy pathway ( 13,22 ). Further, the data reported here were obtained in intestinal epithelial cells that were exposed to lipid solely on their apical surface. Clear differences have been described in the metabolism of FAs absorbed from the basolateral membrane as compared with the apical membrane of this cell type ( 23 ). The proportion of absorbed FA utilized for TAG synthesis versus phospholipid synthesis or conversion to CO 2 is greater when the FA is presented from the apical versus the basolateral side of the membrane ( 24,25 ). Further, studies using cell models yield differing results when performed in whole animals: apoAIV overexpression is associated with increased TAG output by IPEC cells ( 26,27 ) but not in whole animal experiments ( 28 ). Caco2 cells also lack FABP2 that is thought to have a transport role for FA intracellularly ( 29 ). Finally, and most importantly, Caco2 cells appear to require a coat protein II mechanism for TAG to exit the ER ( 30,31 ), whereas rat intestinal cells do not ( 1 ). In sum, these data provide a strong rationale for using whole intestine to study the molecular details of fat absorption.
The most prominent lyso-phospholipid identifi ed in intestinal cytosol was lyso-PC, the majority of which was found in the CEV fraction using chow-fed rats. Ceramide was also considered as a potential activator because of its generation from sphingomyelin by sphingomyelinase located on the brush border of enterocytes ( 36,37 ). Because ceramide concentrations did not vary in the cytosol over a wide range of activity changes in PKC and was found not to be a potent activator of PKC, it is unlikely that ceramide is a physiologically important PKC activator under the conditions of our experiments.
PKC may be activated by three different mechanisms, phosphorylation of T410 by phosphatidylinositol-3-phosphate/P DK-1, autophosphorylation of T560, and a conformational change in the kinase resulting in relief of pseudosubstrate inhibition ( 38 ). In the current experiments, ATP was required for activation, and a conformational change was observed in the kinase on activation by lyso-PC resulting in us being unable to differentiate between autophosphorylation of T560 and the presumed relief of inhibition induced by the pseudosubstrate on binding to lyso-PC.
In the present study, we found that lyso-PC is a physiological activator of PKC . PKC is a member of the PKC superfamily of protein kinases, in particular it is an atypical PKC kinase. The atypical PKC's, specifi cally PKC , are stimulated by a variety of agents such as PS, arachidonic acid ( 32,33 ), phosphatidic acid (PA), lyso-PA, phosphatidylinositol, and phosphatidylglycerol ( 33 ) but not DAG or Ca 2+ . In intestinal cells, we show here that lyso-PC is a potent stimulator over a broad range of activator concentrations. By contrast, in prior experiments lyso-PC has been shown to be a concentrationdependent stimulator/inhibitor of PKC activity ( 32 ), to be a modest stimulator in a nondose ranging study ( 34 ), or to require PS for PKC activity to be expressed ( 35 ).
In this study, we focused on potential stimulators of PKC activity that could be derived from the diet. The major dietary lipids presented for absorption are FAs and MAGs derived from the lipolysis of TAG by pancreatic triacylglycerol lipase and lyso-phospholipids derived from the phospholipolysis of the phospholipids by pancreatic phospholipase A 2 . The most prominent dietary phospholipid is PC, most of which comes from bile ( 19 ).   Fig. 9. Co-immunoprecipitation studies in proteins in cytosol. Bead-bound antibodies to caveolin-1, PKC CD36, and Sar1b were incubated with cytosol isolated from PC-infused rats (1 mg), and the beads were washed to remove unbound proteins. The beads were boiled in Laemmli's buffer, and the proteins were separated by SDS-PAGE. The proteins were transblotted to a nitrocellulose membrane, and the membrane was incubated with the indicated antibodies. Antibody binding was measured by ECL. IgG bound to beads was used as a control.
On activation, PKC is usually considered to move from the cytosol to a specifi c binding site such as tight junctions ( 39,40 ), signaling molecules ( 40 ), or caveolae ( 41 ). By contrast, we show in the intestine that PKC elutes from CEV on activation. One possibility for this is that PKC activation most likely takes place on the surface of the CEV inducing a conformational change in PKC as we show here. This structural alteration results in PKC eluting from the surface of the CEV, enabling it to diffuse in the cytosol quicker and more easily fi nd its substrate, Sar1b, than if it were on CEV. Once bound to Sar1b, it phosphorylates a threonine ( 7 ). The consequence of Sar1b phosphorylation is to free FABP1 from its cytosolic heterotetramer ( 7 ), enabling it to bind to the ER and organize the PCTV budding complex ( 4,6,7 ).
It should be noted that the major actors in the activation of PKC are all present on the surface of CEV. Because of this constricted geographic location, small amounts of activator could potentially have signifi cant effects on its target as we show here with lyso-PC and PKC . Further, the K m of lyso-PC activation of PKC and the measured concentration of lyso-PC in the cytosol in our various models would suggest that in each model, the amount of lyso-PC present would directly affect PKC activation.
The results of our limited proteolysis experiments suggest that PKC bound to CEV is in a more globular form (molten globule) in which many tryptic susceptible hydrolytic sites are hidden. By contrast, on activation by lyso-PC, PKC presumably assumes a more open confi guration resulting in more tryptic attackable sites becoming available.
Our current and past data lead us to propose a new feed forward theory of dietary fat absorption ( Fig. 10 ). This theory would suggest that dietary FAs (MAG) enter the apical membrane of enterocytes via caveolae and are endocytosed as CEVs in proportion to the quantity of absorbed FAs ( 16 ). We propose that the CEVs are targeted to the ER where they unload their FA and MAG (not shown to be present in CEV) cargo. The FAs and MAG are converted to TAG in the ER, in preparation for their incorporation into prechylomicrons. On the surface of CEV is lyso-PC that has been absorbed from the lumen derived mostly from dietary (biliary) PC and PKC . On the surface of the CEV, lyso-PC activates PKC , which elutes from the vesicle, binds to, and phosphorylates Sar1b as a member of an FABP1-containing heteroquatromer in intestinal cytosol ( 7 ). As a complex member, FABP1 cannot bind to the ER ( 7 ). Sar1b phosphorylation disrupts the complex, releasing FABP1, which can now bind to the ER membrane. On binding, FABP1 organizes the PCTV budding complex generating the vesicle that transports the prechylomicron to the Golgi ( 4,6 ). The result of the proposed mechanism is that the ability of the ER to generate and transport prechylomicrons is controlled, at least in part, by the amount of dietary FA delivered to the ER as substrate for Dietary fat enters the enterocyte as FA or MAG and is endocytosed into the cytoplasm via CEV in proportion to the dietary intake. On the surface of the CEV is PKC and lyso-PC. The lyso-PC activates PKC that elutes off the CEV and diffuses into the cytosol. FA (MAG)-containing CEV targeted to the ER where it unloads its FA cargo. The FA and MAG are used as substrates for TAG synthesis, which is incorporated into the prechylomicron. Simultaneously, the activated PKC binds to and phosphorylates Sar1b, which disrupts the Sar1b-FABP1-containing heterodimer. The FABP1 is targeted to the ER where it organizes the PCTV budding complex. chylomicron TAG. This is consistent with the more FAs absorbed, the more CEVs are produced ( 16 ). This would provide a potential mechanism whereby the amount of dietary FA could correlate with PCTV production. Although luminal PC availability also controls chylomicron and PCTV output ( 9,11,13,20,42 ), it should be recognized that the more lipid ingested, the greater the gall bladder contraction and thus PC delivery to the intestine. Thus, the amount of dietary lipid and PC delivery to the intestine is mutually supportive.