Characterization of the Functional Interaction of Adipocyte Lipid-binding Protein with Hormone-sensitive Lipase*

Hormone-sensitive lipase (HSL) is an intracellular lipase that plays an important role in the hydrolysis of triacylglycerol in adipose tissue. HSL has been shown to interact with adipocyte lipid-binding protein (ALBP), a member of the family of intracellular lipid-binding proteins that bind fatty acids and other hydrophobic ligands. The current studies have addressed the functional significance of the association and mapped the site of interaction between HSL and ALBP. Incubation of homogeneous ALBP with purified, recombinant HSL in vitro resulted in a 2-fold increase in substrate hydrolysis. Moreover, the ability of oleate to inhibit HSL hydrolytic activity was attenuated by co-incubation with ALBP. Co-transfection of Chinese hamster ovary cells with HSL and ALBP resulted in greater hydrolytic activity than transfection of cells with HSL and vector alone. Deletional mutations of HSL localized the region of HSL that interacts with ALBP to amino acids 192–200, and site-directed mutagenesis of individual amino acids in this region identified His-194 and Glu-199 as critical for mediating the interaction of HSL with

Hormone-sensitive lipase (HSL) 1 is an intracellular neutral lipase that is highly expressed in adipose and steroidogenic tissues (1). The enzyme has broad substrate specificity, displaying hydrolytic activity against triacylglycerol, diacylglycerol, and cholesteryl ester (2). Observations from HSL-null mice have shown that HSL is responsible for ϳ50% of the neutral triglyceride lipase activity and all of the neutral cholesteryl ester hydrolase activity in white adipose tissue (3). Thus, HSL plays an important role in regulating lipolysis and the release of fatty acids from adipose tissue. The sequence of HSL is unrelated to other mammalian lipases, but it shares sequence and structural similarity with several bacterial and fungal lipases (4 -11). This structural similarity is based on the ability to model a large portion of the C-terminal ϳ450 amino acids of HSL as an ␣/␤ hydrolase (7); however, the initial ϳ320 amino acids of the protein share no sequence or structural homology with any known proteins. Within the C-terminal region of the protein lies a 150-amino acid sequence that contains a number of sites phosphorylated in response to lipolytic stimulation (7,12,13). In this regard HSL is unique among lipases for the ability of its activity to be up-regulated by phosphorylation. In addition to phosphorylation, HSL activity appears to be regulated by oligomerization, with the dimeric enzyme exhibiting markedly increased activity (14).
Utilizing a yeast two-hybrid screen of a rat adipose tissue library, we previously demonstrated that HSL specifically interacts with adipocyte lipid-binding protein (ALBP or aP2) and identified the N-terminal 300 amino acids of HSL as the region responsible for this interaction (15). ALBP is highly expressed in adipose tissue and is a member of the family of intracellular fatty acid-binding proteins that bind fatty acids, retinoids and other hydrophobic ligands (16). It has been proposed that intracellular fatty acid-binding proteins function to sequester fatty acids, thus serving as an intracellular buffer or participating in facilitating the movement of fatty acids within the cell. In view of our observation that HSL and ALBP interact, we proposed that ALBP might prevent feedback inhibition of HSL by high local concentrations of free fatty acids released at the site of hydrolysis. Consistent with this view, adipocytes from ALBP-null mice exhibit markedly reduced basal and stimulated lipolysis both in situ and in vivo (17,18). In the present studies we have addressed the functional significance of the interaction of HSL with ALBP and provide evidence that the interaction of ALBP with HSL constitutes an additional mechanism whereby the hydrolytic activity of HSL is regulated. Furthermore, we have explored the identification of the sequences in HSL that mediate its interaction with ALBP. Construction of Deletional and Mutational Constructs of HSL-Cterminal deletion fragments of HSL were generated by PCR using a common 5Ј-primer (5Ј-GAG AAC CCA CTG CTT ACT) and the following 3Ј-primers: HSL-(1-270), 5Ј-CGG CAA GCT AAG CAG GCG GCT; HSL-(1-240), 5Ј-GGC TTT CCA GAA GTG CAC GTC CAG; HSL-(1-230) 5Ј-TGA TGC GCT CAA ATT CAG; HSL-(1-220), 5Ј-GGT CTA TGG CGA ATC GGC; HSL-(1-209), 5Ј-ACT TGC AGT CAC ACT GAG; HSL- , 5Ј-GTA GTG TTC CCC GAA GGA. Each of the DNA fragments generated from PCR was confirmed by DNA sequencing using an ABI prism DNA sequencer (Applied Biosystems, Foster City, CA). Sitedirected mutagenesis of HSL was carried out using Stratagene "QuikChange" mutagenesis kit. The primer pairs used and the resulting amino acids for different mutants are listed in Table I. The identities of the mutant constructs were confirmed by DNA sequencing using an ABI prism DNA sequencer.

Chemicals and Reagents-All
Production and Purification of Recombinant HSL and ALBP-Recombinant GST-HSL was produced in baculovirus as described previously (14). Recombinant His-HSL was generated by cloning full-length rat HSL cDNA into the SmaI site of pAcHLT-A containing a His 6 tag. pAcHLT-A-HSL (5 g) was co-transfected into Sf21 cells with 0.5 g of BaculoGold TM DNA using the transfection kit from the manufacturer. The titer of the recombinant virus was determined using an end point dilution assay, and the virus was re-amplified to a final titer of 1.5 ϫ 10 7 pfu/ml. To produce recombinant proteins, Sf21 cells were grown in 150-mm Petri dishes and each 2 ϫ 10 7 cells were infected with 100 l of the high titer recombinant virus; cells were harvested 3 days after infection. To metabolically label HSL, Sf21 cells were grown in complete TNM-FH insect medium supplemented with L-[2,3,4,5-3 H]arginine (100 Ci) and L-[4,5-3 H]leucine (100 Ci) for 3 days following infection with baculovirus. After harvesting and cell extraction, His-HSL was purified on a Ni-agarose column. Recombinant ALBP and ALBP mutants were produced in Escherichia coli and purified as described (19,20).
In Vitro Protein-Protein Interaction-After sequence confirmation of the identity of HSL mutants, the HSL mutants were in vitro translated with [ 35 S]methionine by using the TNT ® transcription/translation system (15). GST-ALBP or GST alone were incubated with glutathioneagarose beads in buffer B (20 mM Tris, pH 8.0/0.15 M NaCl/1 mM EDTA/0.5% Nonidet P-40). After 1 h of incubation at room temperature, the beads were washed three times in buffer B and then incubated with [ 35 S]methionine-labeled HSL. After 1 h of incubation at room temperature, the beads were washed five times in buffer B, and proteins that bound to the beads were eluted in SDS/PAGE sample buffer, separated on SDS/10% PAGE, and visualized on a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
Cell Culture and Transfection-CHO cells were grown in Coon's F12/Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum at 37°C under 5% CO 2 . For transient transfection experiments, cells were subcultured at a density of 2 ϫ 10 5 cells/well in 6-well plates the day prior to incubation with 0.75 g of pcDNA3-HSL or pcDNA3-HSL mutants, as well as 0.75 g of pGFP-ALBP and 0.25 g of pCMV ␤-galactopyranoside in 10 l of Lipofectin reagent. Cells were transfected following the procedure from Invitrogen and harvested 40 h after transfection for measurement of HSL activity.
HSL Activity-HSL activity was determined as neutral cholesterol ester hydrolase activity using either a cholesteryl[ 14 C]oleate emulsion for measurement of cellular activity, as described previously (13) Sucrose Gradient Centrifugation-3 H-labeled His-HSL was incubated in the presence of ALBP or GST and then layered on top of a linear 10 -40% (w/w) sucrose gradient (5 ml, prepared in 20 mM Tris-HCl, pH 7.4, and 1 mM EDTA) as described previously (14). The tubes were centrifuged in a SW 50.1 rotor at 50,000 rpm for 4 h. Samples were collected from the bottom of the tube in 20 -22 fractions, and aliquots taken for liquid scintillation counting and determination of sucrose concentration by refractometry.

RESULTS
Since we had previously demonstrated that HSL and ALBP physically interact (15), we sought to determine whether this physical interaction might influence the hydrolytic activity of HSL. As an initial approach, we assayed the effects of increasing concentrations of recombinant ALBP on the ability of baculovirus-produced rat HSL to hydrolyze cholesteryl ester substrate dispersed in ethanol. As shown in Fig. 1A, addition of ALBP caused a dose-dependent increase in cholesteryl ester hydrolytic activity of HSL, which resulted in up to a 2-fold increase in activity at saturating concentrations of ALBP. To explore whether this effect of ALBP on HSL hydrolytic activity was specific for ALBP and whether it depended on the ability of ALBP to bind fatty acids, we compared the effects on HSL activity of a nonspecific protein, GST, and a mutant form of ALBP (R126L, Y128F), which possesses only 1% of the fatty  5Ј-GTG TCC TTC GGG GCC CAC TAC AAA CGC AAC GCG ACG GGC  3Ј-CAC AGG AAG CCC CGG GTG ATG TTT GCG TTG CGC TGC CCG   H194L   5Ј-TCC TTC GGG GAG CTC TAC AAA CGC AAC GCG ACG GGC  3Ј-AGG AAG CCC CTC GAG ATG TTT GCG TTG CGC TGC CCG   E193A/H194L   5Ј-TCC TTC GGG GCC CTT TAC AAA CGC AAC GCG ACG  acid binding capacity of the wild type protein (20), but binds normally to HSL in GST pull down experiments (data not shown). Based upon the ALBP crystal structure (21), the side chains of R126L and Y128F extend into the fatty acid binding cavity. As such, they are not anticipated to participate in protein-protein interactions involving ALBP and HSL. In preliminary experiments we observed that the addition of proteins to the reaction resulted in some nonspecific or surface effects on the interaction of the enzyme with its substrate and caused an increase in activity above that seen in the absence of any other proteins. However, we found that inclusion of insulin, a completely nonspecific protein with regard to fatty acid binding or interaction with HSL (Fig. 1B), in the incubation dampened these nonspecific or surface effects. Thus, insulin (0.2 M) was included in all subsequent assays. As shown in Fig. 1C, the addition of GST caused only a small increase in HSL hydrolytic activity above control. Addition of equimolar amounts of the fatty acid binding mutant of ALBP to the reaction increased HSL activity ϳ2-fold above that seen with GST alone to the level observed with wild type ALBP. These results suggest that under the current conditions the increase in HSL activity seen in the presence of ALBP might be due primarily to its ability to interact with HSL and that the ability of ALBP to bind fatty acid does not seem to be required for this effect. As an independent means of confirming that ALBP can modulate HSL activity, we transiently co-transfected CHO cells with HSL and either vector alone or ALBP and then assayed the cellular extracts for HSL hydrolytic activity. As shown in Fig. 1D, overexpression of ALBP together with HSL led to an approximate 2-fold increase in HSL hydrolytic activity compared with cells transfected with HSL and vector alone. These data demonstrate that the presence of ALBP can lead to higher hydrolytic activity of HSL against its substrate both in vitro and within cells.
We have shown that HSL exists as functional dimers composed of homologous subunits and that dimeric HSL displays dramatically greater activity against cholesteryl ester substrate when compared with monomeric HSL (14). Therefore, to test whether the increase in HSL activity observed in the presence of ALBP was due to a change in the dimerization of HSL, we examined the effects of ALBP on HSL subunit structure. To perform these experiments recombinant His-HSL was metabolically labeled by incubating Sf21 cells, which had been infected with HSL-containing baculovirus, with [ 3 H]arginine and [ 3 H]leucine. Purified 3 H-labeled His-HSL was incubated in the presence of ALBP or GST, and the HSL species were then fractionated by sucrose density gradient centrifugation. As shown in a representative experiment in Fig. 2, two differentsized populations of HSL were observed, consonant with a monomer and dimer; however, no consistent increase in HSL dimerization was seen in the presence of ALBP. Thus, the ALBP-induced increase in HSL activity does not appear to be due to any changes in the dimerization of HSL.
We had originally proposed that the interaction of ALBP with HSL might protect HSL hydrolytic activity from product inhibition by fatty acids released during lipolysis. To examine this directly we assayed residual neutral cholesteryl ester hydrolase activity of purified, recombinant rat HSL to which increasing concentrations of oleate were added in the presence or absence of ALBP. As reported previously (22), addition of oleate resulted in a dose-dependent inhibition of HSL activity (Fig. 3A). The presence of ALBP in the incubation in the absence of any oleate increased HSL activity ϳ50% (Fig. 3B). Furthermore, ALBP attenuated the oleate-induced inhibition of HSL hydrolytic activity, shifting the inhibition to the right. To control for potential nonspecific or surface effects of added proteins on the interaction of HSL with its substrate, we compared the ability of GST, the fatty acid binding mutant of ALBP (mALBP), and native ALBP to prevent oleate-induced inhibition of HSL activity (Fig. 3C). Although some nonspecific effects were observed, ALBP was significantly more effective in preventing inhibition of HSL activity at concentrations up to 25 M oleate. Since the fatty acid binding mutant did not preserve HSL activity as effectively as native ALBP when exposed to higher concentrations of oleate (15-25 M), these data suggest that ALBP protects HSL from product inhibition and that this protective effect is dependent, at least partially, on the ability of ALBP to bind fatty acid.
Having documented that ALBP can modulate HSL hydrolytic activity, we sought to identify the structural determinants within HSL that mediate its interaction with ALBP. Previous studies had initially localized the interaction of ALBP with the 300 N-terminal amino acids of HSL (15). To further define the region of HSL responsible for interacting with ALBP, a series of C-terminal deletions from the N-terminal 300 amino acids of HSL were generated and tested for the ability of the [ 35 S]methionine-labeled in vitro translated products to bind to ALBP. As shown in Fig. 4, [ 35 S]methionine-labeled truncations of HSL and GST alone or ALBP-GST were incubated with glutathioneagarose beads, and the proteins that bound to the beads were washed, eluted, and separated on SDS-PAGE. The truncations to HSL-(1-240), HSL-(1-230), HSL-(1-220), and HSL-(1-209) continued to be able to interact specifically with ALBP-GST, while no interaction was observed with GST alone. However, the truncation to HSL-  showed similar binding to GST alone and ALBP-GST, while HSL-(1-187) failed to bind either GST alone or ALBP-GST. To confirm that the initial 187 amino acids are not involved in interacting with ALBP, a form of HSL was generated that lacked the initial 187 amino acids, HSL-(187-767). This truncated HSL interacted specifically with ALBP-GST, while no interaction was observed with GST alone. From these data it appears that the amino acids between 187 and 209 are responsible for the interaction of HSL with ALBP.
Analyzing the predicted secondary structure of HSL using the Predict Protein PHD prediction program and the GCG prediction program suggested that amino acids 192-200 have the highest probability of being located at the surface of the protein and that they might form a loose or turn structure between two ␤-sheets. Internal deletion of amino acids 191-199 eliminated the binding of HSL to ALBP but also resulted in a protein without hydrolytic activity (data not shown). Likewise, deletion of amino acids 191-196 or 194 -199 also resulted in HSL proteins that failed to bind to ALBP and lacked hydrolytic activity (data not shown). Therefore, we generated a series of single and double mutants of HSL between amino acids 193-199. As shown in Fig. 5, [ 35 S]methionine-labeled mutations of HSL were incubated with GST alone or ALBP-GST and then with glutathione-agarose beads, and the proteins that bound to the beads were washed, eluted, and separated on SDS-PAGE. Mutants E193A, Y195A, Y195D, Y195F, K196V/R197L, and R197A all continued to be able to interact specifically with ALBP-GST, while no interaction was observed with GST alone. In contrast, mutants H194L and E199A failed to bind either GST alone or ALBP-GST, suggesting that these sites are either directly involved in the interaction of HSL with ALBP or that mutation at either site is sufficient to alter the secondary FIG. 3. Interaction of free fatty acids and ALBP on HSL activity. A, the indicated amounts of oleate were added to recombinant GST-HSL and HSL hydrolytic activity was measured as described in Fig. 1. B, the indicated amounts of oleate were added to 5 g of recombinant GST-HSL in the presence or absence of 500 nM ALBP, and HSL hydrolytic activity was measured as described above. C, HSL hydrolytic activity of recombinant His-HSL was measured in the presence or absence of the indicated concentrations of oleate after the addition of insulin (0.  6, 18, 19), pcDNA3-HSL-(187-767) (lanes 7, 20,21) were in vitro translated with [ 35 S]methionine using the TNT ® Transcription/Translation System and incubated with GST (lanes 8, 10,12,14,16,18,20) or ALBP-GST (lanes 9, 11,13,15,17,19,21) and with glutathione-agarose beads. Proteins that bound to the beads were eluted, separated on 10% SDS-PAGE, and visualized using a PhosphorImager ® . structure of HSL in order to eliminate its binding to ALBP.
To assess the functional significance of these mutations, we transiently co-transfected CHO cells with the HSL mutants and an empty vector and then assayed the cellular extracts for HSL hydrolytic activity. As shown in Fig. 6, when co-transfected with vector alone, HSL mutants E193A, H194L, Y195F, R197A, and E199A retained similar hydrolytic activity against cholesteryl ester as wild type HSL; however, HSL mutants Y195A, Y195D, and K196V/R197L resulted in reduced, though measurable, basal hydrolytic activity. Since these mutants retained hydrolytic activity, but mutations at 194 and 199 failed to bind to ALBP, we could now examine whether the specific interaction of ALBP with HSL is required for the ALBP-induced increase in HSL activity. CHO cells were transfected with the mutant HSLs and ALBP, and then assayed for neutral cholesterol esterase activity. Co-transfection of ALBP with wild type HSL increased hydrolytic activity ϳ50%. Likewise, coexpression of ALBP with E193A also increased hydrolytic activity ϳ50%. Co-expression of ALBP with Y195A, Y195D, or K196V/R197L appeared to increase hydrolytic activity ϳ50%; however, because the basal activities of these mutants were reduced, the results should be interpreted cautiously. Interestingly, HSL mutants H194L, Y195F, or E199A, each of which retained normal basal activity, failed to display an increase in hydrolytic activity when co-expressed with ALBP. DISCUSSION Free fatty acids are an essential source of energy for many tissues. The flux of free fatty acids is primarily dependent on the lipolysis of stored triacylglycerol in adipose tissue (23). Multiple mechanisms are involved in controlling lipolysis; however, HSL appears to play a critical role, mediating the initial hydrolysis of triacylglycerol to diacylglycerol, followed by the hydrolysis of diacylglycerol to monoacylglycerol (2). The final step, hydrolysis of monoacylglycerol to glycerol and fatty acid, is mediated by another enzyme, monoacylglycerol lipase (24). One of the mechanisms regulating lipolysis is control of the intracellular localization of HSL, with lipolytic stimulation leading to the translocation of HSL from the cytosol to the lipid droplet in some physiological settings (25)(26)(27). In addition, the hydrolytic activity of the enzyme is increased following phosphorylation by protein kinase A (12,13,28).
Our previous observation that HSL specifically interacts with ALBP led us to propose that ALBP might prevent feedback inhibition of HSL caused by high local concentrations of fatty acids released at the site of hydrolysis (15). An alternate concept would be that ALBP delivers fatty acids to HSL, thereby mediating the inhibition of catalytic activity by lipids.
In the present studies we have provided evidence that the interaction of ALBP with HSL constitutes an additional mechanism whereby the hydrolytic activity of HSL is regulated. Thus, incubation of ALBP with purified, recombinant HSL in vitro resulted in an increase in substrate hydrolysis. This ALBP-induced increase in hydrolytic activity was due to at least two components: first, a small nonspecific effect of added protein, perhaps to alter the surface tension of the substrate and thus its interfacial interaction with the enzyme, and second, a specific effect. The specific effect of the ALBP-induced increase in HSL hydrolytic activity appeared to be due primarily to the ability of ALBP to interact with HSL. The ability of ALBP to bind fatty acid did not seem to be required for this effect, since a fatty acid binding mutant of ALBP, which binds normally to HSL, displayed a similar capacity to increase activity as wild type ALBP. There was no evidence that the ALBP-induced increase in HSL activity was due to any changes in the dimerization of HSL. Furthermore, the ability of ALBP to increase HSL hydrolytic activity was also demonstrated in situ by observing an increase in HSL activity in cells co-transfected with HSL and ALBP as compared with HSL and vector alone. Importantly, the ability of fatty acids to inhibit HSL hydrolytic activity was attenuated by co-incubation with ALBP. This protection of HSL hydrolytic activity was also due both to a nonspecific component and to a specific component. In this case, the specific effect of ALBP to protect HSL from fatty acid induced inhibition appeared to depend in part on the ability of ALBP both to interact with HSL and to bind fatty acids since the ALBP fatty acid binding mutant did not preserve HSL activity as effectively as native ALBP when exposed to higher concentrations of fatty acids. It should be noted that the level of ALBP in vivo is extremely high (estimated to be 250 -400 M), suggesting that, within the cellular context, the ability of ALBP to relieve product inhibition by sequestration of fatty acids is likely to be much greater than measured in vitro or in situ. Indeed, based upon the amount of fatty acids released from isolated adipocytes and the estimated water volume of an adipocyte, the calculated intracellular concentration of fatty acids is estimated to be ϳ600 Ϯ 75 M (29). The near 1:1 stoichiometry of the total fatty acid pool with ALBP and the high affinity of fatty acids for the protein, measured by a combination of titration calorimetry and fluorescence displacement assays, suggest that the fatty acids within the cell are largely, if not exclusively, found protein-bound. Thus, the con- Mutations of HSL were generated as described under "Experimental Procedures." Constructs were in vitro translated with [ 35 S]methionine and incubated with GST or ALBP-GST and with glutathione beads, as described in Fig. 3. Proteins that bound to the beads were eluted, separated on 10% SDS-PAGE, and visualized as described using a PhosphorImager ® .
FIG. 6. Effect of ALBP on hydrolytic activity of HSL mutants. CHO cells were co-transfected with pcDNA3-HSL or pcDNA3-HSL mutants, pGFP-ALBP or pGFP (vector alone), and pCMV ␤-galactopyranoside. Cells were harvested 40 h after transfection for measurement of HSL activity. Results are the mean Ϯ S.E. of triplicate samples and are representative of five independent experiments.