Phosphorylation of hormone-sensitive lipase by protein kinase A in vitro promotes an increase in its hydrophobic surface area


C. Holm, Department of Experimental Medical Science, BMC, C11, SE-221 84 Lund, Sweden
Fax: +46 462224022
Tel: +46 462228581


Hormone-sensitive lipase (EC; HSL) is a key enzyme in the mobilization of fatty acids from stored triacylglycerols. HSL activity is controlled by phosphorylation of at least four serines. In rat HSL, Ser563, Ser659 and Ser660 are phosphorylated by protein kinase A (PKA) in vitro as well as in vivo, and Ser660 and Ser659 have been shown to be the activity-controlling sites in vitro. The exact molecular events of PKA-mediated activation of HSL in vitro are yet to be determined, but increases in both Vmax and S0.5 seem to be involved, as recently shown for human HSL. In this study, the hydrophobic fluorescent probe 4,4′-dianilino-1,1′-binaphthyl-5,5′-disulfonic acid (bis-ANS) was found to inhibit the hydrolysis of triolein by purified recombinant rat adipocyte HSL, with a decrease in the effect of bis-ANS upon PKA phosphorylation of HSL. The interaction of HSL with bis-ANS was found to have a Kd of 1 μm in binding assays. Upon PKA phosphorylation, the interactions of HSL with both bis-ANS and the alternative probe SYPRO Orange were increased. By negative stain transmission electron microscopy, phosphorylated HSL was found to have a closer interaction with phospholipid vesicles than unphosphorylated HSL. Taken together, our results show that HSL increases its hydrophobic nature upon phosphorylation by PKA. This suggests that PKA phosphorylation induces a conformational change that increases the exposed hydrophobic surface and thereby facilitates binding of HSL to the lipid substrate.

Structured digital abstract


adipose triglyceride lipase


4,4′-dianilino-1,1′-binaphthyl-5,5′-disulfonic acid


hormone-sensitive lipase


lipoprotein lipase


protein kinase A




In mammals, fatty acids are mobilized from stored triacylglycerols by the consecutive action of adipose triglyceride lipase (ATGL), hormone-sensitive lipase (HSL), and monoacylglycerol lipase [1]. Phosphorylation of HSL by protein kinase A (PKA) is central to the molecular control of lipolysis, but other events, notably phosphorylation of the lipid droplet protein perilipin, are also of key importance. In adipocytes, stimulation of lipolysis by catecholamines results in activation of adenylate cyclase, leading to elevated levels of cAMP, which causes the catalytic subunits of PKA to dissociate from the regulatory subunits and thereby become active [2,3]. On the other hand, insulin prevents lipolysis, an effect mainly executed via activation of phosphodiesterase 3B, thus lowering cAMP levels. Both HSL and perilipin are phosphorylated directly by PKA, whereas ATGL and its cofactor CGI58 appear to be indirectly controlled by PKA [2,4]. Nevertheless, these phosphorylation events appear to promote the interaction of both ATGL and HSL with the stored lipids, thus increasing hydrolysis of the latter. Whereas CGI58 and perilipin form a complex under basal conditions, they dissociate after phosphorylation of perilipin by PKA. ATGL then forms a new complex with CGI58, rendering ATGL enzymatically active [5,6]. HSL is known to translocate to perilipin-containing lipid droplets after stimulation of lipolysis [7], but a direct interaction between the two proteins has never been proven. The exact molecular events following PKA phosphorylation of HSL and perilipin leading to the activation of lipolysis remain to be elucidated.

Using HSL from several different species, it has been shown that its activity increases approximately 100% after in vitro phosphorylation by PKA [8–10]. In rat HSL, Ser563, Ser659 and Ser660 are phosphorylated by PKA in vitro as well as in vivo [9]. Of these, Ser659 and Ser660, corresponding to Ser649 and Ser650 in human HSL, have been shown to regulate activity in vitro [9,11]. In contrast to the relatively large number of studies devoted to the elucidation of the phosphorylation sites in HSL and the regulation of its translocation, few studies have addressed the effects of PKA phosphorylation on the HSL molecule itself. However, in a recent study, we showed that even though PKA phosphorylation increases the activity of human HSL in vitro, the affinity for triolein (TO) emulsions decreases [9,11]. This may reflect the fact that PKA phosphorylation induces structural changes in the vicinity of the lipid-binding region of HSL. Thus, it is possible that HSL adopts a more open and flexible conformation upon PKA phosphorylation, allowing for easier release of product molecules and leading to an increased turnover rate.

Previous work has shown that lipoprotein lipase (LPL) interacts strongly with, and in fact is inhibited by, the hydrophobic probe 4,4′-dianilino-1,1′-binaphthyl-5,5′-disulfonic acid (bis-ANS) [12]. Another hydrophobic probe, i.e. SYPRO Orange, is now routinely used for differential scanning fluorimetry in thermal denaturation experiments for buffer optimization prior to crystallization trials [13]. As these probes bind to exposed hydrophobic patches in proteins, they were used in this study to generate evidence that PKA-phosphorylated HSL exhibits an increase in the solvent-exposed hydrophobic surface area as compared with the unphosphorylated enzyme. Further proof was obtained from negative stain transmission electron microscopy studies, which demonstrated that HSL interacts more closely with phospholipid vesicles following PKA phosphorylation.


Expression and purification of C-terminally His-tagged rat adipocyte HSL

C-terminally His-tagged rat HSL was successfully expressed in Sf9 insect cells using the baculovirus/insect cell expression system. The protein was purified by anion exchange chromatography followed by nickel affinity chromatography and dialysis (Fig. 1). Western blot analysis confirmed the identity of the purified protein as HSL (data not shown). The yield of pure protein was 3 mg per litre of insect cell culture. The specific activities of the purified protein against TO, 1-mono-oleoyl-2-O-mono-oleylglycerol and cholesterol oleate were 2.6 U·mg−1, 30 U·mg−1, and 3.5 U·mg−1, respectively. These specific activities are lower than those previously reported for nontagged recombinant rat HSL, but are in accordance with the published activity of His-tagged human HSL [11,14].

Figure 1.

 Purity of recombinant HSL. SDS/PAGE gel displaying the expression and purification of recombinant C-terminally His-tagged rat adipocyte HSL. Lane 1: supernatant fraction of lysed Sf9 cells expressing HSL. Lane 2: molecular mass marker. Lane 3: purified rat adipocyte HSL.

HSL inhibition by bis-ANS

To evaluate whether bis-ANS had an effect on the enzymatic activity of HSL, lipase assays were performed with increasing amounts of bis-ANS using TO as substrate. Bis-ANS inhibited the activity of both phosphorylated and nonphosphorylated HSL. Normalization of the two sets of data revealed differences between the normalized activities of phosphorylated and nonphosphorylated HSL at bis-ANS concentrations above 5 μm, indicating that the bis-ANS interaction with HSL is altered upon phosphorylation of HSL by PKA (Fig. 2).

Figure 2.

 Inhibition of HSL lipase activity by bis-ANS. HSL, phosphorylated or nonphosphorylated, was preincubated for 10 min and assayed in the presence of the given concentrations of the hydrophobic probe bis-ANS, using TO as substrate. The activity of phosphorylated HSL in the TO assay was normalized to the activities of nonphosphorylated HSL, and the activities were compared (unpaired nonparametric t-test, n = 4).

HSL interaction with bis-ANS

To evaluate the binding of bis-ANS to nonphosphorylated HSL, we measured the fluorescence of bis-ANS in complex with HSL. With an excitation wavelength of 296 nm, the emission was scanned between 300 nm and 550 nm at bis-ANS concentrations ranging from 0.1 to 10 μm, both with and without added HSL. The HSL–bis-ANS complex fluorescence was derived by subtracting spectra of bis-ANS alone from spectra obtained with added HSL. The HSL–bis-ANS complex maximum emission wavelength ranged from 476 nm at 0.1 μm bis-ANS to 488 nm at 10 μm bis-ANS, indicating that there could be more than one binding site for bis-ANS on HSL (Fig. 3A). The maximum emission intensity of the complex increased in an inverse hyperbolic fashion (r2 = 0.99 for an inverse hyperbolic curve fit), and Kd for the complex was determined to be 1.00 μm bis-ANS (Fig. 3B).

Figure 3.

 Interaction of bis-ANS with HSL. Spectra of bis-ANS obtained at concentrations ranging from 0.1 to 10 μm were subtracted from spectra of bis-ANS mixed with HSL in order to generate difference spectra displaying the interaction between HSL and bis-ANS (A). The maximum fluorescence increased according to the concentration of bis-ANS, following a hyperbolic curve (r2 = 0.998), which provided a Kd of 1.0 μm for the HSL–bis-ANS complex (B). The spectra shown have been smoothed (using two neighbouring values).

HSL phosphorylation and activation with variable ATP concentrations

Because ATP interacted with both bis-ANS and SYPRO Orange, creating high levels of background fluorescence, we investigated the possibility of using lower amounts of ATP for the phosphorylation of HSL. Using radiolabelled ATP, we showed that, by using only 15 μm ATP in the phosphorylation reaction mix, we could obtain a similar extent of phosphorylation to that obtained at an ATP concentration of 200 μm (Fig. 4A). The stoichiometry of phosphorylation was 0.20 mol phosphate per mol HSL for the 200 μm ATP reaction, in accordance with results published for human HSL [11], and 0.16 mol phosphate per mol HSL for the 15 μm ATP reaction. In accordance with the similar degree of phosphorylation at the two ATP concentrations, there was no significant difference between the activation levels obtained at the two different ATP concentrations with TO as substrate (Fig. 4B). Incubatation of the enzyme preparations with alkaline phosphatase did not affect the enzymatic activity (data not shown). This indicates that HSL purified from the baculovirus/insect cell system was obtained in a dephosphorylated form, at least with regard to activity-controlling sites. This is in agreement with previous reports for His-tagged human HSL and non-tagged rat HSL [11,15].

Figure 4.

In vitro phosphorylation and activation of HSL. HSL was phosphorylated in the presence of radiolabelled ATP, with total ATP concentrations in the phosphorylation reaction of 15 μm or 200 μm, and analysed for incorporation of 32P (A) and activity against the TO substrate (B). The results in (B) include two individual experiments (n = 6). Data represents means ± standard error of six assays. *P < 0.05, ***P < 0.0005, unpaired, nonparametric t-test.

HSL interaction with bis-ANS and SYPRO Orange after phosphorylation

To investigate whether HSL gains hydrophobic surface area upon phosphorylation by PKA, we analysed the interaction of phosphorylated HSL with bis-ANS in comparison with nonphosphorylated HSL, using fluorescence. Even at ATP concentrations of 15 μm in the phosphorylation reaction mix, resulting in a final concentration below 150 nm in the fluorescence measurements, the interaction between bis-ANS and ATP was too strong for reliable spectra to be obtained. Therefore, HSL samples were dialysed after phosphorylation and then reanalysed for protein content before being used in fluorescence measurements. Spectra of samples containing phosphorylated and dialysed HSL mixed with bis-ANS were recorded, and spectra of dialysed phosphorylation mixes (including PKA, which provided only minor contributions to the total fluorescence) lacking HSL were subtracted to eliminate the influence of interactions with buffers and PKA, thus providing difference spectra solely representing the interaction between HSL and bis-ANS (Fig. 5A,B). The spectra for phosphorylated HSL were above the spectra for nonphosphorylated HSL for all three concentrations tested.

Figure 5.

 Comparison of fluorescence from the hydrophobic probes bis-ANS and SYPRO Orange in complex with phosphorylated and nonphosphorylated HSL. HSL was phosphorylated by PKA, dialysed, and mixed with bis-ANS, and spectra were recorded at an excitation wavelength of 296 nm. Spectra of reaction mixes containing no HSL were subtracted to generate the displayed difference spectra illustrating the interaction between bis-ANS and HSL. The concentrations of bis-ANS used were 1.5 μm and 2.0 μm in (A) and (B), respectively. HSL was phosphorylated with 15 μm ATP in the reaction mix, and mixed with SYPRO Orange, and spectra were recorded at an excitation wavelength of 492 nm. Spectra of reaction mixes lacking HSL or both HSL and kinase were subtracted from the spectra of the nonphosphorylated and phosphorylated HSL samples, respectively, to generate the displayed difference spectra illustrating the interaction between SYPRO Orange and HSL. The concentrations of SYPRO Orange used were × 0.25 and × 0.5 in (C) and (D), respectively. The spectra shown are smoothed (using four neighbouring values) and normalized to the maximum fluorescence of the complex between phosphorylated HSL and the respective probes.

Owing to the interaction between bis-ANS and ATP, the binding assay provided only reliable performance in a short range of the bis-ANS concentrations tested in Fig. 2, and only 1 μm, 1.5 μm and 2 μm bis-ANS provided acceptable signal-to-noise ratios. At low bis-ANS concentrations, the HSL–bis-ANS complex fluorescence was lost in noise, and at high concentrations, the absolute fluorescence signal was out of range for the instrumentation. Thus, to further verify the increase in hydrophobicity of HSL after phosphorylation by PKA, we also applied an alternative hydrophobic probe, i.e. SYPRO Orange. HSL was incubated with or without PKA, and the fluorescence after excitation at 492 nm was recorded by scanning the emission between 550 nm and 800 nm. In order to create difference spectra illustrating solely the interaction between HSL and SYPRO Orange, spectra of phosphorylation reaction mixes lacking both HSL and PKA or lacking only HSL were subtracted from the recorded spectra of the nonphosphorylated or phosphorylated samples, respectively. An advantage with this approach was that even though ATP also interfered with these measurements (the performance of the assay was reliable only within a limited range of probe concentrations), dialysis after phosphorylation was not necessary, probably because of a lower degree of interaction of ATP with SYPRO Orange than with bis-ANS. The molar concentration of SYPRO Orange is impossible to calculate, as the molecular mass is not publicly available, but the relative concentrations of SYPRO Orange tested in the measurements were × 0.2, × 0.25, × 0.5, and × 1. For all tested concentrations of SYPRO Orange, the fluorescence of phosphorylated HSL lay above the fluorescence of nonphosphorylated HSL, indicating that HSL gains hydrophobic surface area upon phosphorylation by PKA (Fig. 5C,D).

Electron microscopy of phosphorylated and nonphosphorylated HSL

As a first attempt to investigate whether the increase in hydrophobic surface area is reflected by increased binding of phosphorylated HSL to lipid surfaces, we investigated the interaction between HSL and phospholipid vesicles using negative stain electron microscopy. Phospholipid vesicles mimic the lipid droplets found in vivo, which are covered by a single layer of phospholipids, but avoid the problems of lipid hydrolysis during analysis, as HSL is known not to exhibit phospholipase activity. Furthermore, we know from previous work that HSL associates with phospholipid vesicles [15]. Thus, sonicated phosphatidylcholine vesicles were mixed with either nonphosphorylated or PKA-phosphorylated HSL, stained with uranyl formate, and analysed by transmission electron microscopy. In the obtained micrographs, HSL appeared as light particles even when observed inside the vesicles. When vesicles mixed with nonphosphorylated HSL (Fig. 6A) were compared with vesicles mixed with phosphorylated HSL (Fig. 6B), 12% of the imaged vesicles contained nonphosphorylated HSL, and 82% of the imaged vesicles contained phosphorylated HSL (based on observing 300 vesicles for each condition). In addition, the content of HSL in each vesicle was markedly increased for phosphorylated HSL (Fig. 6D) when compared with nonphosphorylated HSL (Fig. 6C), reflecting a stronger interaction with phospholipids and/or the fact that phosphorylated HSL more easily penetrates the phospholipid membrane to gain access to the underlying lipid substrate. The presence of HSL in the vesicles was confirmed by immunogold electron microscopy (Fig. 6E,F). When vesicles mixed with nonphosphorylated HSL (Fig. 6E) were compared with vesicles mixed with phosphorylated HSL (Fig. 6F), 23% of the imaged vesicles contained nonphosphorylated HSL, and 74% of the imaged vesicles contained phosphorylated HSL (based on observing 300 vesicles for each condition), which is in good agreement with the observations made without immunogold labelling (Fig. 6A–D).

Figure 6.

 Negative stain electron microscopy analysis of the interaction between HSL and phospholipid vesicles. Electron micrograph illustrating HSL integrated into phospholipid vesicles in either the nonphosphorylated form (A, C) or phosphorylated form (B, D). HSL appears as white shadows inside vesicles to a much higher extent, and is present in higher numbers per vesicle for the phosphorylated enzyme than for the nonphosphorylated one. Immunogold labelling confirms the presence of HSL associated with and inside the vesicles [nonphosphorylated HSL in (E), and phosphorylated HSL in (F)]. The size bar in (F) corresponds to 50 nm (A, B, E, F) and 25 nm (C, D).


In this study, we used recombinant rat adipocyte HSL produced in Sf9 cells using baculovirus-mediated expression to demonstrate that HSL undergoes a conformational change upon PKA phosphorylation, which increases the solvent-exposed hydrophobic surface area.

The fluorescent probe bis-ANS has previously been used to study the lipid-binding properties of another mammalian lipase, i.e. LPL [12]. Bis-ANS was found to bind tightly to LPL in the vicinity of the active site and also to inhibit the enzymatic activity [12]. Similarly, we show here that bis-ANS inhibits the lipase activity of HSL. Interestingly, there were only minor effects of bis-ANS concentrations up to 60 μm on HSL activity when preincubation of the enzyme with bis-ANS prior to assaying was omitted. We believe that this is due to the presence of BSA in the assay buffer and the lipid substrate emulsion, but not in the preincubation buffer, in accordance with the observation that LPL was not inhibited by bis-ANS when BSA was present in the assay. This is also supported by the observation that even though the HSL–bis-ANS complex has a Kd of 1 μm (Fig. 3), much higher concentrations are needed to decrease HSL activity (Fig. 2). Scavenging of bis-ANS by BSA is presumably the reason for a sigmoidal inhibition curve, rather than the expected hyperbolic one.

The binding of bis-ANS to HSL followed an inverse hyperbolic saturation curve. The estimated Kd of 1 μm is lower than what has been reported for most other proteins [16], indicating that HSL exhibits high affinity for bis-ANS, although not as high as that of LPL, for which the Kd was reported to be 0.10–0.26 μm [12]. The maximum emission wavelength of the HSL–bis-ANS complex shifted 13 nm from the lowest to the highest concentrations of bis-ANS. This shift suggests that there is more than one binding site for bis-ANS on HSL.

The change in solvent-exposed hydrophobic surface area of HSL following PKA phosphorylation was examined using both bis-ANS and SYPRO Orange. Because of the interaction of both bis-ANS and SYPRO Orange with ATP, we were forced to use significantly lower ATP concentrations in these experiments than those normally used. Thus, prior to the fluorescence experiments with these hydrophobic probes, we established that the use of 15 μm ATP in the phosphorylation reaction resulted in the same degree of both phosphorylation and activation as the use of 200 μm ATP (Fig. 4A,B). The interaction between bis-ANS and dialysed phosphorylated HSL was measured using 1 μm, 1.5 μm and 2 μm bis-ANS. The fluorescence of the phosphorylated form of HSL was found to be substantially higher than that of the nonphosphorylated form at all three concentrations. This result was verified using SYPRO Orange, which interacts to a lesser degree with ATP than does bis-ANS. This enabled us to employ a simplified protocol without the need for dialysis prior to fluorescence measurements when using only 15 μm ATP in the preceding phosphorylation reactions. Relative concentrations of × 0.2, × 0.25, × 0.5 and × 1.0 SYPRO Orange were used for the comparison of phosphorylated and nonphosphorylated HSL. The fluorescence from phosphorylated HSL was higher than that from nonphosphorylated HSL for all four concentrations of SYPRO Orange, thus strengthening the argument that HSL gains solvent-exposed hydrophobic surface area upon phosphorylation. Electron microscopy of phospholipid vesicles mixed with phosphorylated and nonphosphorylated HSL demonstrated a more pronounced interaction with the vesicles for the phosphorylated variant, in terms of both the number of vesicles invaded by HSL and the larger HSL content of the individual vesicles (Fig. 6). This may be due to the increased hydrophobic nature of phosphorylated HSL as compared with nonphosphorylated HSL, although alternative explanations exist. For instance, it is possible that phosphorylated HSL binds more avidly to the polar head of the phospholipids and that this is followed by an interaction between the apolar acyl chains of the phospholipids and side chains of particular amino acids, thus accounting for the increased capacity to penetrate to the interior of the vesicle. Phospholipid vesicles mimic the lipid droplets found in vivo, but avoid the problem of hydrolysis, as HSL lacks phospholipase activity. It is indeed possible that binding of phosphorylated HSL to the phospholipid vesicles, followed by penetration of the membrane, mimics what happens in vivo as HSL is anchored to the lipid droplet to hydrolyse acylglycerols.

Even though PKA-phosphorylated HSL was found to bind more bis-ANS than nonphosphorylated HSL, the inhibition of TO activity by bis-ANS was decreased upon phosphorylation. A possible explanation for this apparent discrepancy could be that PKA phosphorylation induces a conformational change that increases the accessible hydrophobic surface area, enabling freer access to the lipid-binding site, in return for weaker binding. This is well in line with our recent kinetic measurements on human HSL, showing that PKA phsophorylation increases both maximum turnover rate and S0.5 [11].

In this study, we provide evidence that HSL gains accessible hydrophobic surface area upon PKA phosphorylation. This gain in hydrophobic surface area presumably accounts for the increase in in vitro activity of HSL following PKA phosphorylation through increased binding between HSL and the lipid substrate emulsion. It is possible that the gain in accessible hydrophobic surface area not only affects the ability of HSL to interact with the lipid droplet, but is also is involved in driving the translocation of HSL that occurs upon lipolytic stimulation of adipocytes [7]. The exact molecular events involved in the translocation of HSL are, as yet, incompletely understood. It seems clear that perilipin is required for translocation of HSL, although a direct interaction between the two proteins has been not been demonstrated [5,6,17]. Data are emerging that point to perilipin as a key player in directing proteins involved in lipolysis to a subset of lipid droplets [5]. Interestingly, we recently showed that the affinity of human HSL for TO decreased in in vitro asssays upon PKA phosphorylation [11]. Taken together with the results presented here, this underscores the fact that the affinity measured in activity assays involves several aspects of lipase activity, i.e. adsorption, entry and binding of individual lipid molecules to the enzyme.

Future studies will be needed to determine whether the phosphorylation-induced gain in hydrophobic surface area described here affects other properties of HSL than binding to lipids, e.g. binding to lipid droplet-associated proteins.

In conclusion, our results demonstrate that HSL increases its hydrophobic nature upon phosphorylation by PKA. Thus, it can be speculated that phosphorylation of HSL by PKA induces a conformational change that exposes and/or increases the lipid-binding area of the enzyme. A direct demonstration of this presumed conformational change will have to await the solving of the atomic structure of HSL in its native and phosphorylated forms.

Experimental procedures

Expression and purification of C-terminal His-tagged recombinant rat adipocyte HSL

To generate a recombinant baculovirus encoding C-terminally tagged rat adipocyte HSL, full-length rat adipocyte HSL cDNA, including a sequence encoding one Pro residue and eight C-terminal His residues, was subcloned into the BamHI and XbaI sites of pVL1393, as follows. The PCR product obtained using the sense primer 5′-ATC ATC TCC ATC GAC TAC TCC CTG-3′, the antisense primer 5′-AAG AAT TCT AGA TTA ATG GTG ATG ATG GTG ATG ATG GTG TGG GGT CAG CGG TGC AGC AGG GGG GGT-3′ (XbaI sites underlined; His-tag in italic) and pVL1393–HSL [18] as template was digested using XbaI and BssHII and subcloned into pVL1393–HSL. PCR was performed using Vent polymerase (New England Biolabs, Ipswich, MA, USA), and the PCR product was sequenced using BigDye (Applied Biosystems, Foster City, CA, USA) upon subcloning. Recombinant virus was generated by transfecting Sf9 cells using the BaculoGold Transfection Kit (BD Biosciences Pharmingen, San Diego, CA, USA), according to the manufacturer’s instructions but using the Sf-900 II medium instead of TMN-FH. Plaque purification was performed and high-titre virus stocks were generated using standard procedures.

For protein expression, Sf9 insect cells were grown at 27 °C in suspension cultures (160 r.p.m.) in Sf-900 medium supplemented with 4% fetal bovine serum and 1% penicillin/streptomycin (all from Gibco, through Invitrogen AB, Lidingö, Sweden). Cell cultures (2 × 106 cells·mL−1) were infected at a multiplicity of infection of 10. Infection was followed by a 72 h expression period. Cells were harvested by centrifugation (1200 g, 10 min), and resuspended in five pellet volumes of lysis buffer (50 mm Tris/HCl, pH 8.0, 1 mm dithiothreitol, 1 mm EDTA, 1% C13E12, 10% glycerol). The cell suspension was gently sonicated and centrifuged for 45 min at 4 °C and 50 000 g. The supernatant fraction was filtered through a 0.22 μm filter and loaded onto a Q-Sepharose Fast Flow anion exchange column (GE Healthcare, Uppsala, Sweden). The column was washed with 10 volumes of 50 mm NaCl, 20 mm Tris/HCl (pH 8.0), 1 mm dithiothreitol, 1 mm EDTA, 0.01% C8E4, and 10% glycerol, and then eluted with approximately two column volumes of 1 m NaCl, 20 mm Tris/HCl (pH 8.0), 0.1 mm dithiothreitol, 0.01% C8E4, and 10% glycerol.

Protein eluted from the Q-Sepharose column was loaded directly onto a nickel affinity chromatography column (Ni2+–nitrilotriacetic acid Superflow; Qiagen, Valencia, CA, USA), washed with 10 volumes of 18 mm imidazole, 300 mm NaCl, 50 mm Tris/HCl (pH 8.0), 0.1 mm dithiothreitol, 1% Triton X-100, and 10% glycerol, and 15 volumes of 5 mm imidazole, 300 mm NaCl, 50 mm Tris/HCl (pH 8.0), 0.1 mm dithiothreitol, 0.01% C8E4, and 10% glycerol, and eluted with a stepwise gradient towards 250 mm imidazole, 300 mm NaCl, 50 mm Tris/HCl (pH 8.0), 1 mm dithiothreitol, 0.01% C8E4, and 10% glycerol. The eluted protein was then dialysed overnight against 50 mm Tris/HCl (pH 8.0), 300 mm NaCl, 1 mm dithiothreitol, 0.01% C8E4, and 10% glycerol, and stored at −80 °C. Protein amounts were measured by the 2D Quant method (GE Healthcare) and the Bradford method [19]. The latter underestimated HSL content by a factor of 1.5 relative to the 2D Quant method. The C-terminally His-tagged rat HSL was used for all analyses in this study except for the electron microscopy studies, where nontagged rat HSL, expressed and purified as described in [15], was used.

HSL activity assays

HSL lipase activity was measured against phospholipid-stabilized emulsions of TO, 1-mono-oleoyl-2-O-mono-oleylglycerol or cholesterol oleate [18,20]. Briefly, labelled and nonlabelled lipid substrates and phosphatidylcholine/phosphatidylinositol (3 : 1) in cyclohexane solutions were dried under a stream of N2, and this was followed by emulsification by sonication and addition of 2% BSA (1-mono-oleoyl-2-O-mono-oleylglycerol assay) or 5% BSA (TO and cholesterol oleate assays). Enzymes were diluted to a suitable concentration in 100 μL of 20 mm potassium phosphate (pH 7.0), 1 mm EDTA, 1 mm dithiothreitol, and 0.02% BSA, and 100 μL of the emulsified substrate was added and mixed. Reactions were typically incubated for a period of 30 min at 37 °C before the reaction was quenched by the addition of 3.25 mL of methanol/chloroform/heptane (10 : 9 : 7) and 1.1 mL of 0.1 m potassium carbonate and 0.1 m boric acid (pH 10.5). Samples were then vortexed and centrifuged (800 g, 20 min), and the content of released fatty acids in the upper phase was determined by scintillation counting. For all assays, we confirmed that the reaction velocity was constant during the 30 min incubation period.

HSL inhibition by bis-ANS

HSL was incubated for 5–10 min in 50 mm Tris/HCl (pH 8), 300 mm NaCl, 10% glycerol, 1 mm dithiothreitol, 0.02% C8E4 and 10 mm MgCl2 containing either 5 μm, 10 μm, 15 μm, 20 μm, 30 μm, 45 μm or 60 μm bis-ANS at room temperature and assayed against TO as previously described, with the exception that the resulting reaction mixtures contained either 5 μm, 10 μm, 15 μm, 20 μm, 30 μm, 45 μm or 60 μm bis-ANS. For all assays including bis-ANS, the reaction rates were constant for at least 30 min of incubation at 37 °C.

HSL phosphorylation using 32P-labelled ATP

HSL (9 μg) was phosphorylated at room temperature in 110 μL volumes containing 50 mm Tris/HCl (pH 8), 300 mm NaCl, 10% glycerol, 1 mm dithiothreitol, 0.02% C8E4, 10 mm MgCl2, 25 U of PKA (New England Biolabs), protease inhibitor cocktail (Roche Complete; Roche Diagnostics, Mannheim, Germany), and either 200 μm ATP and 0.3 μCi·μL−1 [32P]ATP[γP], or 15 μm ATP and 0.0225 μCi·μL−1 [32P]ATP[γP]. Aliquots were taken after 4, 8, 16, 32 and 64 min of incubation, and quenched by the addition of Laemmli buffer [21]. In control reactions, PKA was omitted. Samples were analysed by SDS/PAGE, stained with Coomassie, scanned, and slab dried. 32P-labelled HSL was detected as described above. For quantification of incorporated phosphate, HSL bands were excised from the gel and placed in scintillation vials containing 10 mL of scintillation liquid and quantified on a scintillation counter (Wallac 1414 liquid scintillation counter; Perkin Elmer, Waltham, MA, USA). The original reaction mixtures were included as standards.

HSL phosphorylation by PKA for activity and hydrophobicity measurements

HSL (4 μg) was phosphorylated in 50 μL volumes containing 50 mm Tris/HCl (pH 8), 300 mm NaCl, 10% glycerol, 1 mm dithiothreitol, 0.01% C8E4, 10 mm MgCl2, 15 μm or 200 μm ATP and 0.25 U of PKA/μL supplemented with a protease inhibitor cocktail (Roche Complete) for 1 h at room temperature. For activity measurements, the protein was diluted to suitable amounts and assayed in the TO assay. For hydrophobicity measurements using SYPRO Orange, 8 μL of the phosphorylation reaction mixture was used in measurements.

Fluorescence measurements including HSL and bis-ANS

Fluorescence measurements were performed essentially as in [12]. One millilitre of 50 mm Tris/HCl (pH 8), 300 mm NaCl, 10% glycerol, 1 mm dithiothreitol, 0.01% C8E4, and 10 mm MgCl2, including the respective concentration of bis-ANS, was excited at 296 nm, and emission was scanned from 300 to 550 nm. Subsequently, samples of HSL in 50 mm Tris/HCl (pH 8), 300 mm NaCl, 10% glycerol, 1 mm dithiothreitol, 0.01% C8E4 and 10 mm MgCl2 were added to the cuvette, and fluorescence was recorded similarly. The first spectrum not containing HSL was subtracted from the later spectra containing HSL, creating difference spectra reflecting the interaction between bis-ANS and HSL. The amount of HSL used for each spectrum ranged between 0.8 μg and 2.4 μg, depending on the concentration of bis-ANS.

Owing to a strong interaction between bis-ANS and ATP, phosphorylated HSL samples had to be dialysed twice for 2 h against 10 000 volumes of 50 mm Tris/HCl (pH 8), 300 mm NaCl, 10% glycerol, 1 mm dithiothreitol, 0.01% C8E4 and 10 mm MgCl2 to remove ATP and thereby decrease background fluorescence. After dialysis, samples were centrifuged for 20 min at 25 000 g, and protein contents were remeasured using the Bradford method before analysis. Spectra were recorded as described above. Reactions without added HSL were used as controls for the interaction between bis-ANS and PKA: spectra recorded with samples containing only PKA were subtracted from spectra containing HSL. Even after dialysis of samples, there was a considerable interaction with ATP, and signal-to-noise ratios decreased from 5.7 without ATP added to 1.4 in samples including ATP.

Fluorescence measurements including HSL and SYPRO Orange

For the evaluation of the increase in hydrophobicity of HSL after phosphorylation by PKA, the commercial probe SYPRO Orange (from Molecular Probes through Invitrogen AB) was used. The measurements were performed in quartz cuvettes by adding samples to 1 mL of 50 mm Tris/HCl (pH 8), 300 mm NaCl, 10% glycerol, 1 mm dithiothreitol, 0.01% C8E4, 10 mm MgCl2, and SYPRO Orange. The sample was excited at 492 nm, and emission was scanned from 550 to 800 nm.

HSL was phosphorylated as described above, using 15 μm ATP in the reaction mix. In parallel with the phosphorylation reaction, a mock phosphorylation reaction without added PKA was used for measurements on nonphosphorylated HSL. Similarly, a reaction in which HSL was replaced by dialysis buffer and without PKA was performed, and a spectrum of this reaction was subtracted from the spectrum recorded for the nonphosphorylated HSL sample. The experiments were carried out at concentrations of SYPRO Orange ranging from × 0.25 to × 1. The concentrations providing the best reproducibility for these experiments were × 0.25 and × 0.5, where the signal-to-noise ratio was the highest, i.e. 2.4 in the absence of PKA, and 3.2 in the presence of PKA.

Negative stain transmission electron microscopy analysis of the interaction between HSL and sonicated phospholipid vesicles

Phosphatidylcholine vesicles were prepared as described in [15]. In brief, phosphatidylcholine was evaporated under nitrogen to remove the solvent. Evaporation was repeated twice after addition of 0.1 mL of freshly distilled, dried diethyl ether. The resulting lipid film was placed under reduced pressure for 12 h, and then allowed to swell for 30 min at room temperature in 20 mm Tris/HCl (pH 7.0), 0.1 m NaCl, 1 mm EDTA and 1 mm dithioerythritol at a final concentration of 25 mg·mL−1 phosphatidylcholine. After swelling, the solution was sonicated with 0.5 s pulses for 45 min at 4 °C under nitrogen with a microtip sonicator (model B-15P; Branson, Danbury, CT, USA) at a setting of 50% of maximum intensity. The clear solution obtained was centrifuged at 100 000 g for 60 min to remove multilamellar vesicles and titanium from the microtip. Vesicle samples were mixed with HSL and immediately prepared for electron microscopy. In some experiments, HSL containing vesicles were incubated for 30 min at room temperature with antibodies against HSL that were conjugated with 5 nm colloidal gold as described by Baschong and Wrigley [22]. All protein concentrations were in the 10–20 nm range. Subsequently, 5 μL aliquots of the solution were adsorbed onto carbon-coated grids for 1 min, washed with two drops of water, and stained on two drops of 0.75% uranyl formate. Prior to this, the grids were rendered hydrophilic by glow discharge at low pressure in air. Specimens were observed in a Jeol JEM 1230 electron microscope operated at 60 kV accelerating voltage (Jeol, Tokyo, Japan). Images were recorded with a Gatan Multiscan 791 CCD camera (Gatan UK, Abingdon, UK) [23].


We would like to thank B. Danielsson and M. Baumgarten for excellent technical assistance, and R. Wallén and E. Hallberg (Cell and Organism Biology, Lund University) for help with electron microscopy. Financial support was obtained from the Swedish Research Council (project no. 11284 to C. Holm, and project no. 7480 to M. Mörgelin), the Swedish Diabetes Association, Faculty of Medicine at Lund University, and the following foundations: Novo Nordisk, A. Påhlsson, Salubrin/Druvan, Johan och Greta Kock, Alfred Österlund, Crafoord, Konung Gustav V:s 80-årsfond and Torsten and Ragnar Söderberg. C. Krintel was supported by the Research School in Pharmaceutical Sciences (FLÄK).