Secretion of triacylglycerol-poor VLDL particles from McA-RH7777 cells expressing human hepatic lipase.

Hepatic lipase (HL) plays a role in the catabolism of apolipoprotein (apo)B-containing lipoproteins through its lipolytic and ligand-binding properties. We describe a potential intracellular role of HL in the assembly and secretion of VLDL. Transient or stable expression of HL in McA-RH7777 cells resulted in decreased (by 40%) incorporation of [3H]glycerol into cell-associated and secreted triacylglycerol (TAG) relative to control cells. However, incorporation of [35S]methionine/cysteine into cell and medium apoB-100 was not decreased by HL expression. The decreased 3H-TAG synthesis/secretion in HL expressing cells was not attributable to decreased expression of genes involved in lipogenesis. Fractionation of medium revealed that the decreased [3H]TAG from HL expressing cells was mainly attributable to decreased VLDL. Expression of catalytically-inactive HL (HLSG) (Ser-145 at the catalytic site was substituted with Gly) in the cells also resulted in decreased secretion of VLDL-[3H]TAG. Examination of lumenal contents of microsomes showed a 40% decrease in [3H]TAG associated with lumenal lipid droplets in HL or HLSG expressing cells as compared with control. The microsomal membrane-associated [3H]TAG was decreased by 50% in HL expressing cells but not in HLSG expressing cells. Thus, expression of HL, irrespective of its lipolytic function, impairs formation of VLDL precursor [3H]TAG in the form of lumenal lipid droplets. These results suggest that HL expression in McA-RH7777 cells result in secretion of [3H]TAG-poor VLDL.

scription Kit and the iQ SYBR Green Supermix were purchased from Qiagen (Mississauga, ON, Canada) and BioRad (Mississauga, ON, Canada), respectively.

Construction of expression plasmids
The expression plasmid encoding human HL (pCMV5-HL) was generated as previously described ( 29 ). The plasmid encoding the catalytic inactive HL SG (pCMV5-HL SG ), in which the active site Ser-145 was mutated into Gly (S145G), was generated by site-specifi c mutagenesis using the QuikChange mutagenesis kit (Stratagene). The sense oligonucleotide used for mutagenesis was 5 ′ -CCTAATTGGGTACAGCCTGGGTGC-ACACGTGT-CAGG-3 ′ . The adenovirus vector encoding HL was prepared using the AdEasy adenoviral system (Q-Biogene, Carlsbad, CA) according to manufacturer's instructions.

Cell culture and transfection
McA-RH7777 cells were obtained from the American Type Culture Collection and cultured in DMEM containing 10% FBS and 10% horse serum. Stable cell lines were generated by cotransfection with pSV2neo together with either pCMV5-HL or pCMV5HL SG using Lipofectamine (Invitrogen) or calcium phosphate precipitation ( 30 ) methods, respectively. The stable clones were selected with G418 (400 g/ml). After screening for HL expression, the stable clones were maintained in G418 (200 g/ ml). For adenovirus mediated HL expression, cells were incubated with HL-encoding adenovirus (40 pfu per cell) in a minimal amount of serum-free DMEM (1 ml per 60 mm dish) for 1 h. After infection, the cells were cultured in normal media for an additional 36 h prior to experiments.

HL activity assay
Stable cell lines expressing HL or HL SG were cultured for 4 h with serum-free DMEM supplemented with heparin (100 U/ml). The conditioned medium was used as the enzyme source to determine HL activity using [ 14 C]tributyrin as a substrate according to described protocols ( 31 ). The amount of [ 14 C]butyrate released into the aqueous phase was quantifi ed by scintillation counting.

Immunoblot analysis
The cells were incubated with serum-free DMEM medium ± heparin (100 U/ml) for 4 h. The cells were harvested in sample loading buffer (SLB: 10 mM Tris, pH 8.0, 8 M urea, 2% SDS, 10% glycerol, 5% ␤ -mercaptoethanol, and bromophenol blue), and the secreted HL proteins were adsorbed onto hydrated fumed silica (Cab-o-sil) as previously described ( 32 ) and eluted in SLB. The cell and medium protein samples were loaded onto polyacrylamide gels based on equal cell protein levels and transferred to nitrocellulose membrane for immunoblot analysis with appropriate antibodies against HL and actin.

Metabolic labeling of lipids and proteins
For lipid labeling, cells were labeled for the indicated times with [ 3 H]glycerol (5 Ci/ml) in DMEM containing 20% FBS and 0.4 mM oleate and either with or without 100 U/ml heparin. Total lipids were extracted from cells and media and resolved by TLC as previously described ( 33 ) ( 14,15 ). On the other hand, overexpression of human HL in HL-null/ LDLR-null background results in ‫ف‬ 60-70% reduction of plasma VLDL-associated TAG, PL, and cholesterol under chow or Western diet ( 16,17 ). Liver-specifi c expression of human HL in HL-null/apoE-null mice also results in ‫ف‬ 40% reduction in plasma levels of TAG, PL, and cholesterol ( 18 ). These in vivo animal studies have fi rmly established a role for HL in regulating plasma apoB-containing lipoproteins; however, it remains unclear whether the HL action increases or decreases the risk of developing atherosclerosis ( 6 ). Although the majority of the data describes the extracellular role of HL in the catabolism of circulating lipoproteins, some studies have suggested that HL may possess lipolytic activity intracellularly (19)(20)(21). The intracellular activity of HL has been detected within the endoplasmic reticulum (ER)/Golgi secretory pathway in transfected Chinese hamster ovary cells ( 20 ). In the present study, we tested the hypothesis that expression of HL in hepatic cells may attenuate the assembly and secretion of lipid-rich VLDL. The process of VLDL assembly is initiated during translation and translocation of apoB-100 across the ER membrane ( 22 ). The nascent VLDL particle is further enlarged in TAG content through a "second step" lipidation process, where bulk TAG is incorporated to form mature VLDL. In McA-RH7777 cells, maturation of TAG-rich VLDL is achieved in post-ER compartments ( 23 ), and the TAG utilized for VLDL maturation is present in the microsomal lumen in the form of lipid droplets ( 24,25 ). Formation of these lumenal lipid droplets (LLD) appears to require the activity of microsomal triglyceride-transfer protein (MTP) (26)(27)(28). Recent experimental evidence has suggested that formation of LLD under lipid-rich conditions may also require apoC-III ( 24,25 ). However, how LLD-TAG is incorporated into the nascent VLDL precursor is not known, nor is it clear about lipid or protein factors that regulate the formation of LLD. Data obtained from the present transfection studies suggested that transient or stable expression of HL, regardless of its catalytic activity, exerted a negative impact on the formation of LLD and, consequently, the assembly and secretion of TAG-rich lipoproteins under lipid-rich conditions.  35 S-label into cell-associated or medium apoB-100 was not diminished and, rather, was increased in HL expressing cells, although the increase in [ 35 S]apoB-100 did not reach statistical signifi cance ( Fig. 2A ). Synthesis or secretion of [ 35 S]albumin, another hepatic protein used as a control, was unaffected by HL expression ( Fig. 2B ). The decreased [ 3 H]TAG secretion without accompanying decrease in [ 35 S]apoB-100 secretion provide the fi rst hint for secretion of lipid-poor VLDL particles upon HL expression.

Decreased [ 3 H]TAG secretion as VLDL upon HL expression was independent of the catalytic function of HL
We next determined whether or not the decreased [ 3 H] TAG secretion upon HL expression was attributable to the immunoprecipitated from the cells and media and resolved by SDS-PAGE (5% gel for apoB; 12% gel for albumin and actin). Radioactivity associated with the 35 S-labeled proteins was quantifi ed by scintillation counting.

Subcellular fractionation
Cells were labeled with [ 3 H]glycerol for the indicated times, homogenized, and fractionated as previously described ( 23 ). The postnuclear supernatant was subjected to 165,000 × g for 30 min to isolate the microsomal pellet. The membranous and lumenal fractions of the microsome were separated as previously described ( 23 ). Lipids were extracted from all the fractions and analyzed by TLC as described above.

Density fractionation of lipoproteins by cumulative rate fl otation ultracentrifugation
Cells were labeled with [ 3 H]glycerol or [ 35 S]methionine/ cysteine for the indicated times as described above. Lipoproteins, either present in the media or within the microsomal lumen, from metabolically labeled cells were fractionated into VLDL 1 (S f >100), VLDL 2 (S f 20-100), and other lipoproteins as described previously ( 27 ). Lipids were extracted from each fraction and analyzed by TLC as described above.

Real-time RT-PCR
Total RNA was extracted using TriZol reagent following the manufacturer's instructions. Reverse transcription was performed using QuantiTect Reverse Transcription Kit according to manufacturer's directions. PCR was performed on the iCycler (BioRad) using iQ SYBR Green Supermix following manufacturer's instructions. The cycle threshold values were normalized to cyclophilin A ( Ppia ). Sequences of the PCR primers are listed in supplementary Table I.

Statistical analysis
Values were expressed as mean ± SD. The signifi cance of differences between control and HL expressing cells or between control and HL SG expressing cells were analyzed using Student's t -test.

Expression of hepatic lipase resulted in decreased synthesis and secretion of [ 3 H]TAG
In the current studies, the potential role of HL on hepatic TAG-rich lipoprotein assembly and secretion was determined by transient or stable expression of wild-type or the catalytically-inactive form of HL in McA-RH7777 cells. Among stable clones that expressed different levels of HL (supplementary Fig. I A), the TAG hydrolysis activity (assayed using tributyrin as a substrate) was readily detectable in heparin-treated media (supplementary Fig. I B), confi rming that the recombinant enzyme was catalytically active. A very low amount of HL was detectable in the conditioned media containing no heparin (data not shown), suggesting that the recombinant HL, as expected, was mainly bound to extracellular matrix ( 29 ). With increasing levels of HL expression in the stable clones (

Expression of HL, irrespective of catalytic activity, results in attenuated formation of lumenal lipid droplets rich in [ 3 H]TAG
The above data suggest that HL expression might have exerted an effect on VLDL assembly and secretion. To gain an insight into mechanisms underlying the HL action intracellularly, we determined the effect of HL expression on the formation of LLD that are precursors for TAG-rich VLDL maturation during the second step lipidation ( 24,25 ). Metabolic control cells ( Fig. 5A ). However, accumulation of [ 3 H]TAG in LLD was reduced (by ‫ف‬ 40%) catalytic activity of the enzyme. To this end, we prepared stable cell lines expressing a mutant HL in which Ser-145 at the catalytic site was substituted with Gly (designated HL SG ) (supplementary Fig. IV A). As expected, the mutant HL SG secreted from the transfected cells lost its TAG hydrolysis activity (supplementary Fig. IV B). However, metabolic labeling experiments showed that expression of the mutant HL SG also resulted in decreased [ 3 H]TAG secretion (by >50%), the effect similar to that of wild-type HL ( Fig.  3A , open bars). Fractionation of lipoproteins showed that the decreased [ 3 H]TAG secretion from HL and HL SG cells was mainly attributable to decreased VLDL 1 and VLDL 2 (by ‫ف‬ 75% and ‫ف‬ 60%, respectively) relative to control cells ( Fig. 3B , top panel). These results indicate that expression of HL, regardless of its catalytic activity, decreased VLDL-TAG secretion. Secretion of [ 3 H]PC as VLDL 1 or VLDL 2 from HL or HL SG cells was unchanged as compared with control cells ( Fig. 3C , open bars; Fig. 3D , top panel).
It has been suggested previously ( 34-36 ) that HSPGanchored HL can capture the secreted TAG-rich lipoproteins on the cell surface. To determine whether or not the decreased VLDL-[ 3 H]TAG from HL or HL SG cells was attributable to surface-anchored HL, we performed metabolic labeling in the presence of heparin to prevent HL cell surface binding. Under this condition, the amount of [ 3 H]TAG recovered from the HL or HL SG media was increased as compared with heparin-free conditions ( Fig.  3A , closed bars). This result indicated that a signifi cant amount of newly secreted VLDL particles was indeed captured on the surface of cells expressing either HL or HL SG . However, as compared with neo control, the released [ 3 H] tween microsomal membranes and microsomal lumen ( Fig. 6C ) ( Fig. 6C ) suggested that HL might play a role in TAG partitioning between microsomes and cytosol. Indeed, the relative distribution of [ 3 H]TAG showed pronounced presentation in cytosol and reduced association with microsomes in HL or HL SG expressing cells relative to neo controls ( Fig. 6D ). As mentioned earlier ( Fig. 1B ) ( Fig. 6B ), as well as be-

DISCUSSION
The present results provide in vitro evidence that expression of the HL protein, not necessarily the HL activity, results in the secretion of TAG-poor VLDL particles ( Fig.  3 ) with no effect on apoB-100 ( Fig. 4 ). Secretion of the TAG-poor VLDL particles from cells expressing HL (regardless of the catalytic activity) is most likely the consequence of impaired second step lipidation of VLDL particles, which results from the diminished formation of TAG precursors in the form of LLD ( 24,25 ) ( Fig. 5A ). The decreased TAG precursor pool within the microsomal lumen in HL expressing cells is not entirely attributable to impaired TAG synthesis. Rather, expression of HL

Altered expression of lipogenesis genes in cells expressing HL or HL SG
Finally, we determined expression of genes involved in VLDL assembly/secretion (e.g., Apob and Mttp ), lipogenesis (e.g., Dgat1 and Dgat2 ), and ␤ -oxidation (e.g., Cpt1a ) in cells expressing HL or HL SG . As shown in Fig. 7A , expression of HL or HL SG did not impair, but rather stimulated, the apoB and Mttp mRNA. These results are in accord with the metabolic labeling data indicating uncompromised apoB synthesis and secretion in cells expressing HL or HL SG ( Figs. 2, 4 ). The level of Dgat1 mRNA was unchanged and Dgat2 mRNA increased signifi cantly in HL or HL SG expressing cells as compared with that in neo controls ( Fig. 7B ). The relative expression level of Cpt1a , the gene encoding carnitine palmitoyl transferase involved in ␤ -oxidation, was also increased signifi cantly as compared with neo controls ( Fig. 7B ). Thus, expression of HL or HL SG in McA-RH7777 cells did not compromise  crease in secreted VLDL-TAG mass and apoB-100 ( 37 ). Thus, expression of triacylglycerol hydrolase promotes TAG-rich VLDL secretion through increased TAG substrate availability, which is probably achieved via the process termed hydrolysis/re-esterifi cation. Recently, evidence has been obtained for another ER-associated protein, namely arylacetamide deacetylase, whose expression resulted in decreased [  ( 38 ). On the other hand, expression of Esterase-x had no effect on [ 3 H]TAG or apoB-100 secretion, even though TAG synthesis was impaired ( 39 ). The present study is the fi rst demonstration that HL, a secretory lipase, also negatively impacts TAG-rich VLDL assembly/secretion by attenuating the availability of TAG substrates without affecting apoB-100 secretion. Decreased secretion of TAG with no concomitant decrease in apoB-100 has also been observed in vivo in animal models such as liver specifi c inactivation of the transcription factor XBP1 in mice ( 40 ).
Transfection studies have shown that HL possessed catalytic activity in the ER ( 20 ) and Golgi ( 19 ), albeit the intracellular activity was 10 times lower than that of the secreted enzyme. Intracellular maturation of HL involves conversion of catalytically inactive mass into functional, dimeric conformation prior to secretion ( 20 ), a slow process that occurs in the ER and takes hours to accomplish ( 41 ). The prolonged residence time of HL within the ER/Golgi secretory pathway may exert a negative effect on mobilization of TAG into microsomal lumen for VLDL assembly. The mechanism by which the HL traversing the ER/Golgi secretory pathway affects TAG partitioning into microsomes is currently unclear. However, the present mutagenesis study indicates that the catalytic activity of HL is not absolutely required for this effect. A group of HL-interacting proteins have been identifi ed that may play a role in intracellular HL maturation process, none of which, however, appear to be involved in lipid binding or partitioning ( 41 ). Lipid binding regions are present within HL molecules ( 42 ). It has been shown recently that the fully functional homodimer of HL is associated with a microsomal membrane protein ( 43 ). Thus, it is tempting to speculate that (irrespective of catalytic activity) resulted in attenuated partitioning of newly synthesized TAG into the microsomes, particularly into the microsomal lumen ( Fig. 6 ). These data add another level of complexity to the multifunctional HL in hepatic lipoprotein metabolism. Thus, in addition to the well-documented extracellular role of HL in catalyzing TAG/PL hydrolysis and facilitating lipoprotein clearance through receptor-or HSPG-mediated endocytosis, HL may also play an intracellular role in the assembly and secretion of TAG-rich VLDL.
The mechanism by which HL, a secretory hydrolytic enzyme that traverses through the ER/Golgi pathway, affects TAG metabolism intracellularly, thereby infl uencing VLDL assembly and secretion remains to be defi ned. However, many microsome-resident protein/enzyme factors have been identifi ed that play a role in mobilization and utilization of TAG for VLDL assembly and secretion (37)(38)(39). Expression of these microsome-associated proteins/ enzymes has been shown to exert different impacts on the metabolism of hepatic TAG and apoB-100 associated with VLDL. For instance, cells expressing triacylglycerol hydrolase exhibited increased depletion of storage TAG (with little effect on [ 3 H]TAG synthesis) and accompanied in-   the assembly and secretion of TAG-rich lipoproteins, and the mode of action of HL is independent of its hydrolytic activity. microsomal traversing HL may interact with the membranes directly or indirectly (via other lumenal proteins) thereby affecting partitioning of newly synthesized TAG. Early studies have shown that partitioning of TAG into microsomal lumen for VLDL assembly/secretion requires the activity of MTP ( 26 ). Although it is unclear whether or not HL would affect the MTP activity, determination of the Mttp mRNA suggested that its abundance was not diminished in HL expressing cells.
Several hepatic proteins, including apoE and HL, have been shown to play a role in the process termed secretionrecapture that regulates, extracellularly, the net output of apoB-containing lipoproteins from the liver ( 36 ). The present study confi rmed that the recombinant HL expressed in McA-RH7777 was vividly bound to the cell surface in a heparin-sensitive manner. By comparing the respective levels and lipoprotein distribution of [ 6A ). On the other hand, real-time RT-PCR analysis indicated that expression of genes involved in TAG synthesis was undiminished ( Fig. 3B ), suggesting mechanisms other than synthesis might be involved. Enhanced expression of Cpt1a may suggest a role for ␤ -oxidation ( Fig. 3B ), but how HL expression could lead to enhanced ␤ -oxidation is unclear. What remains to be determined is whether the intracellular catalytic activity of HL can be regulated by other protein factors. A potential regulation of extracellular HL activity by angiopoietin-like proteins, a family of secreted glycoproteins, has been reported ( 44 ).
In summary, the present cell culture studies have revealed an intracellular role of HL expression in attenuating