Structure-Activity Studies of Human Sterol Carrier Protein 2"

Recombinant human sterol carrier protein 2 (SCP2) variants were generated by site-directed mutagenesis and expression in Escherichia coli. The ability of the variants to stimulate microsomal conversion of 7-dehy-drocholesterol to cholesterol (sterol carrier activity) and to transfer cholesterol and phosphatidylcholine from donor small unilamellar vesicles to acceptor mem- branes (cholesterol and phosphatidylcholine transfer activities) was compared with wild-type recombinant SCP2. Our results indicate that all measured activities of recombinant human pre-SCP2 (including the 20-amino acid leader sequence) and mature SCPP were similar. Expressed glutathione S-transferase fusion proteins (GST-SCP2 and GST-pre-SCP2) possessed consider- able activity, suggesting that steric obstruction at the amino terminus causes only minor inactivation. The ef- fect of progressive removal of peptides from the carboxyl terminus showed that amino acids between Lys1O0 and Asn'" are essential for SCPB activity. This conclusion was substantiated by the observation 3' sequence 5'-GGAATI'CC-3' (EcoRI site) with 5' primer 3. The resulting DNA fragments were double digested with BamHI and EcoRI and subcloned in BamHI-EcoRI digested vector pGEX-2T. Specific mutagenesis of DNA fragments was performed by PCR am-plification of cDNA encoding human pre-SCP2 (pBS-hSCW17) with (31). The following substitutions were introduced: Leuzo to Glu (ClT to 21-mers introducing specific substitutions as described by Higuchi et al. GAA), Asp70 to Asn (GAC to AAC), Cys71 to Val (TGC to GTI') and to Ser (TGC to TCT), Asnlo4 to Asp (AAC to GAC), Asn104 to Ile (AAC to ATC), and Metlo6 to Leu (ATG to CTG). The resulting DNA fragments were subcloned in pGEX-2T as described above. Specificity of each mutagenesis was verified by direct double-stranded DNA sequencing of the pGEX-2T-containing insert. The double mutant AsplO.vlos, which was also used for activity studies, resulted from a spontaneous mutation in codon 106 (GGT to GAT; Gly*06 to Asp) detected by nucleotide sequencing of a clone presumed to encode the Asplo4 variant. hrification

Recombinant human sterol carrier protein 2 (SCP2) variants were generated by site-directed mutagenesis and expression in Escherichia coli. The ability of the variants to stimulate microsomal conversion of 7-dehydrocholesterol to cholesterol (sterol carrier activity) and to transfer cholesterol and phosphatidylcholine from donor small unilamellar vesicles to acceptor membranes (cholesterol and phosphatidylcholine transfer activities) was compared with wild-type recombinant SCP2. Our results indicate that all measured activities of recombinant human pre-SCP2 (including the 20amino acid leader sequence) and mature SCPP were similar. Expressed glutathione S-transferase fusion proteins (GST-SCP2 and GST-pre-SCP2) possessed considerable activity, suggesting that steric obstruction at the amino terminus causes only minor inactivation. The effect of progressive removal of peptides from the carboxyl terminus showed that amino acids between Lys1O0 and Asn'" are essential for SCPB activity. This conclusion was substantiated by the observation that replacing Asn'" with Asp or Ile caused considerable inactivation, whereas replacing Met' " with Leu had almost no effect. Since N-ethylmaleimide is known to inhibit SCPP activity, substitutions were also introduced in the vicinity of Cys7'. Whereas Val'' and Ser7' variants possessed wild-type activity, replacing Asp70 with Asn almost completely abolished SCP2 activity. Further, the importance of residues located close to the amino terminus was indicated by complete inactivation of a 10-amino-terminal amino acid deletion mutant and by replacing Leu20 with Glu. Circular dichroism results showed that Leu20 and Asp7' may serve to stabilize the overall fold, whereas residue 104 appears to play a role in the specific lipid binding andor transfer activity of SCP2.
Sterol carrier protein 2 (SCP2,I or nonspecific lipid transfer protein) has repeatedly been suggested to participate in the intracellular transport of cholesterol and phospholipids (for review see Ref. 1). It was shown that SCPS activates enzymatic ' The abbreviations used are: SCP2, sterol carrier protein 2; hSCP2, lymerase chain reaction; wt, wild-type; SUV, small unilamellar vesicle.

837-225.
conversion of 7-dehydrocholesterol to cholesterol by liver microsomes (2). The protein also stimulates acyl-CoA cholesterol acyltransferase-mediated esterification of intracellular cholesterol (3-5) and the introduction of less polar substrates in bile acid biosynthesis to membrane-bound enzymes in vitro (6,7). The protein may be required for intracellular transfer of cholesterol, which is needed for pregnenolone synthesis in adrenals (8-10) and ovaries (11). In addition, SCPB promotes the exchange of a wide variety of lipids and sterols between membranes in vitro (5, [12][13][14]. Nucleotide sequencing of SCPB cDNAs revealed that rat, mouse, and human SCPS is synthesized as a 143-amino acid precursor that is processed to the 123-amino acid mature SCPB (16)(17)(18)(19)(20)(21). In addition, it was shown that livers of all studied mammalian species contain SCP2-related transcripts encoding larger proteins that are identical at their carboxyl-terminal domains with pre-SCP2 (16)(17)(18)(19)(20)(21)(22)(23)(24). The complete cDNA encoding one of these proteins (SCPx) has a n open reading frame of 547 codons (18,19,, representing an extension of 404 codons at the initiator methionine of pre-SCP2. Recent identification of an oleic acid-inducible SCPB homologue in Candida tropicalis (25) indicates that SCP2 is highly conserved during evolution and may be of key importance for cellular lipid metabolism.
However, knowledge about the structural basis for SCP2mediated lipid transport is incomplete. Consistent with the apparent absence of extended hydrophobic domains, purified SCPS is water-soluble and does not contain any bound lipid (8,26). In addition, a stable association between SCPS and phospholipids or cholesterol could not be demonstrated (26,27). Therefore, it was proposed that the protein either binds lipids with low affinity (28) or does not function as a typical binding protein but facilitates lipid exchange between membranes by providing a hydrophobic bridge or tunnel between two lipid interphases (35). To map domains of SCPS which are involved in its sterol carrier and lipid transfer activities we constructed mutant proteins that were assayed for their ability to stimulate microsomal conversion of 7-dehydrocholesterol to cholesterol (sterol carrier activity) and to transfer cholesterol and PC from donor SUVs to acceptor vesicles (cholesterol and PC transfer activities).
Deletion-and Site-directed Mutagenesis-Progressive deletions of residues were introduced from the amino and carboxyl termini of SCP2 by PCR deletion mutagenesis with the cDNAencoding human pre-SCP2 (clone pBS-hSCPW17) as template. Oligonucleotide primers (27-mers) including the 5' sequence 5'-CGGATCC-3' (BamHI site) were selected for successive deletions of pentapeptides from the amino terminus and were combined with 3' primer 1 for PCR amplifications. Progressive deletions of 8.18, and 23 amino acids from the carboxyl terminus were produced accordingly by combining 27-mers including the 3' sequence 5'-GGAATI'CC-3' (EcoRI site) with 5' primer 3. The resulting DNA fragments were double digested with BamHI and EcoRI and subcloned in BamHI-EcoRI digested vector pGEX-2T.
Specific mutagenesis of DNA fragments was performed by PCR amplification of cDNA encoding human pre-SCP2 (pBS-hSCW17) with (31). The following substitutions were introduced: Leuzo to Glu (ClT to 21-mers introducing specific substitutions as described by Higuchi et al. GAA), Asp70 to Asn (GAC to AAC), Cys71 to Val (TGC to GTI') and to Ser (TGC to TCT), Asnlo4 to Asp (AAC to GAC), Asn104 to Ile (AAC to ATC), and Metlo6 to Leu (ATG to CTG). The resulting DNA fragments were subcloned in pGEX-2T as described above. Specificity of each mutagenesis was verified by direct double-stranded DNA sequencing of the pGEX-2T-containing insert. The double mutant AsplO.vlos, which was also used for activity studies, resulted from a spontaneous mutation in codon 106 (GGT to GAT; Gly*06 to Asp) detected by nucleotide sequencing of a clone presumed to encode the Asplo4 variant. according to (29,30). Purified fusion proteins were dialyzed against 100 volumes of phosphate-buffered saline and cleaved by a 2-h incubation at 37 "C with Y20 (w/w) thrombin (Sigma). Glutathione S-transferase was removed by affinity chromatography over glutathione-Sepharose 4B (Pharmacia LKB Biotechnology Inc.). Recombinant sterol carrier proteins were dialyzed against 100 volumes of 50 m~ NaC1, 50 m~ Tris-HC1, pH 8.6, and were purified further to homogeneity via passage over DEAE-Sephadex equilibrated with the same buffer. Wild-type sterol carrier proteins and variants were recovered from the flow-through fraction. Protein concentrations were routinely determined with the Bradford assay (52). Sterol Carrier and Lipid Dansfer Assays"SCP2 activity was determined essentially as described previously (2,32). The assays contained microsomes (2 mg), 100 nmol of 7-dehydrocholesterol (Sigma), and 1.2 m~ NADPH (Boehringer Mannheim) in a total volume of 1 ml. Tests containing 1.5-5 mg of rat liver cytosol were used as positive controls. The assays were routinely incubated in a rotary shaker at 37 "C under an atmosphere of argon for 90 min. Thereafter, the reaction tubes were transferred to ice, and the reactions were stopped by adding 1 ml of freshly prepared 15% (w/v) ethanolic KOH solution. Lipids were extracted with n-hexane (Merck, Darmstadt, F. R. G.) and 7-dehydrocholesterol was quantified by recording the W spectrum between 280 and 320 nm. Sterol carrier activity was calculated by subtracting the AEzavszoobtained without SCP2 from the bE292/320 ,,,,, measured in the presence of SCP2 or the variants.
Cholesterol and PC transfer activities were measured by monitoring SCP2-mediated transfer of cholesterol or PC from donor SUVs to acceptor Bacillus megaterium protoplasts prepared as described in (46). For the preparation of SWs, 8 mg of lipid containing 40 mol % cholesterol (Sigma) and 60 mol % soybean PC (Sigma) was dissolved in 1 ml of CHClJmethanol (Wl), and the SUVs were produced essentially as described in (47,48) except that multilamellar vesicles were removed from the preparation by gel filtration over Sepharose 4B (Pharmacia). Transfer assays contained protoplasts (2.5 mg of protein) and liposomes (160 nmol of cholesterol or 160 nmol of soybean PC) in 0.5 ml of SPA buffer (0.3 M sucrose, 0.3% (w/v) NaN3, 0.06 M potassium phosphate, pH 6.2). The assay mixtures were preincubated at 37 "C for 5 min, and samples containing various amounts of SCP2 or variant proteins were added in a volume of 200 pl of 15 m~ potassium phosphate, pH 6.2. m e r 30 min, protoplasts were separated from SUVs by centrifugation at 8,000 rpm with an Eppendorf 5415C centrifuge for 4 min. Protoplasts were washed three times with 500 pl of ice-cold SPA buffer and were then resuspended in 100 pl of 2% Triton X-100,0.45 m~ CaCl,, 50 m~ Tris-HC1, pH 7.8. m e r three freeze-thawing cycles, cholesterol and PC were quantified as described previously (49,50).

RESULTS AND DISCUSSION
In most eukaryotic cells, the majority of free cholesterol resides in the plasma membrane, whereas intracellular membranes contain almost no cholesterol (36). Conversely, the receptor-mediated uptake of exogenous lipoproteins and de novo cholesterol synthesis generate significant amounts of cholesterol in membranes of the endoplasmic reticulum (37), peroxisomes (38, 391, and lysosomal compartments (for review see Ref. 40). These findings led to the proposal of mechanisms facilitating the movement of cholesterol from certain intracellular membranes to the plasma membrane and the sites of intracellular cholesterol utilization.
The hypothesis that SCP2 participates in this process is based upon the findings that the protein catalyzes in vitro transport of a wide variety of sterols (41) and most common lipids (for review see Ref. 35). On the other hand, purified SCPB does not contain bound lipid (8,261, and a stable association between SCPS and radiolabeled sterols could not be demonstrated (41). Therefore, it was proposed that SCP2 does not function as a typical binding protein but facilitates lipid exchange between membranes by other means (35). In this hypothesis, the protein is bound to the surface of the membrane via electrostatic interactions where it functions as a tunnel or bridge between two lipid interphases, thereby facilitating movement of hydrophobic transport substrates from a donor to an acceptor membrane. This model would explain the apparent lack of tight substrate specificity which is observed in all in vitro SCP2 transport assays. However, the results of fluorescence studies suggested low affinity binding of A5,7,9(11)*22ergostatetraen-3/3-01 with a 1:l stoichiometry and a K d of 1.2-1.6 p~ (28), indicating that the lipid-SCP2 complex is an important intermediate in intermembrane sterol transfer. In addition, time-resolved and steady-state fluorescence studies did not present evidence for the immobilization of SCP2 at the membrane surface. These studies showed instead that SCPB shields bound lipids from the aqueous phase in the manner of a typical lipid carrier protein (42).
In view of these two conflicting mechanistic models of SCPBmediated lipid transport we used site-directed mutagenesis to contribute to a better understanding of specific domains and amino acid residues of SCP2 which are involved in its lipid transfer activity. An important prerequisite for successful employment of this approach was the efficient prokaryotic expression of different forms of active SCPB and constructed variants. We show that the pGEX expression system (29,30), in which proteins are expressed as glutathione S-transferase fusion proteins after induction with isopropyl-l-thio-P-D-galactopyanoside, can be used for high level expression of a wide variety of SCPS forms including rat and human pre-SCP2 and SCPB, as well as several deletion variants and substitution mutants. Fig.  1 summarizes the purification procedure for recombinant hSCP2. The presence of a protein band migrating according to the expected size of the SCP2-GST fusion protein (40.5 kDa) is indicated by an arrow in Fig. 1, lane 1 1. Expression and purification of recombinant hSCP2. A 250-ml culture of E. coli XL-1 Blue transformed with vector pGEX-2T/ hSCP2 was grown to an absorbance of -0.8 (600 nm). High level expression of GST-SCP2 fusion protein was induced with 0.1 rn isopropyl-l-thio-P-~-galact~pyranoside, and the culture was incubated further on a rotary shaker at 37 "C for 5 h. Extracts were prepared and recombinant hSCP2 was purified as described under "Materials and Methods." Aliquots of M O O of the total samples were taken from each step of the purification procedure and were subjected to electrophoresis on a 12.5% sodium dodecyl sulfate-polyacrylamide gel as described in (43). protein consisted of close to 50% of total soluble proteins of recombinant bacteria. Essentially pure fusion protein was obtained after a single passage over glutathione-Sepharose 4B (lane 2). The specificity of thrombin digestion is indicated in lane 3. It is shown that thrombin cleaved the fusion protein in two peptides of 26.0 kDa (glutathione S-transferase) and 14.5 kDa (hSCP2). The glutathione S-transferase peptide was removed by passage over glutathione-Sepharose 4B (lane 41, and the remaining thrombin-derived contaminants were removed by ion-exchange chromatography over DEAE-Sephadex, resulting in a homogeneous preparation of recombinant SCP2 (lane 5). This procedure was also successfully performed on human pre-SCP2 (including the 20-amino acid leader sequence), rat pre-SCP2 and SCP2, and all deletion and substitution variants of SCPS which were constructed for this work (not shown).
The absence of mutations within the constructs was verified by DNA sequencing and by automated Edman degradation of the first 20 amino acids showing that because of the introduction of the BamHI linker sequence at the fusion point during the cloning procedure, recombinant human SCP2 and pre-SCPS contained the additional amino-terminal dipeptide Gly-Ser. Mass spectrometry performed with purified recombinant hSCP2 provided a molecular mass of 13,383 2 1.4 Da (calculated mass 13,386) (data not shown).
Isolated recombinant SCPS was highly active in PC and cholesterol transfer as well as in in vitro stimulation of the microsomal conversion of 7-dehydrocholesterol to cholesterol (Fig. 2 and Table I previously from rat liver (2). Recombinant glutathione S-transferase, however, which was purified from bacteria transformed with the pGEX-2T construct without cloned insert, was inactive (Table I), suggesting that the glutathione S-transferase part of the fusion protein did not contribute to the activity. The absence of activity in this case was also interesting because glutathione S-transferase had been described previously as a high affinity sterol-binding protein (44). The activity of SCF'2 also did not differ from that of recombinant pre-SCP2 (Table I), including the 20-amino acid leader sequence, which is absent from mature SCP2 (33, 34). The transfer of cholesterol or PC from SUVs to B. megaterium protoplasts was plotted against the amount of added SCP2, pre-SCP2, GST-SCP2, or GST-pre-SCP2 in Fig. 2 -terminal deletions N-5 and N-10  correspond to variants N-A5-hSCP2 and N-A10-hSCP2, whereas carboxyl-terminal deletions C-8, C-18, and C-23 correspond to variants C-A8-hSCP2, C-Al8-hSCP2, and C-A23-hSCP2. Asterisks indicate that the results are based on measurements with fusion protein.
were virtually identical. Similar results were also reported recently for rat SCPS and pre-SCP2 (51). Therefore, it is likely that intracellular removal of the prepeptide does not lead to substantial activation of SCP2.
Expressed fusion proteins GST-SCP2 and GST-pre-SCP2 also possessed considerable specific sterol carrier activities which, if calculated on a molar basis, ranged between 30 and 50% of purified recombinant hSCP2 (Table I). The maximal specific cholesterol and PC transfer activities of the fusion proteins were 62 pmol of cholesteroVpmo1 of proteidh and 80 pmol of PC/pmol of proteidh (Fig. 2). Thus, the fusion proteins had molar cholesterol and PC transfer activities of -30% as compared with isolated recombinant hSCP2. These results indicate that the addition of more than 200 heterologous amino acids at the amino terminus causes only minor steric inhibition of SCP2-mediated sterol carrier, PC, and cholesterol transfer activities. We believe that these results appear to be hardly compatible with a mode of action involving the whole molecule binding to the surface of a donor membrane followed by dimerization with a second SCPB molecule bound to the acceptor membrane, thereby providing a bridge or tunnel for lipid transfer as was suggested in (35). Our results instead favor the existence of a local lipid binding site. However, as GST-SCP2 and GST-pre-SCP2 fusion proteins in general possessed lower molar activities than SCP2 itself, it is possible that the residues added to the amino terminus of SCP2 exert a partial local effect or shield a putative binding site thereby causing some inhibition. This assumption would imply that residues located in the vicinity of the amino terminus would significantly contribute to its activity. To study this question we progressively removed the pentapeptides from the amino terminus of SCP2 and measured the sterol carrier activity of the deletion variants.
Although the deletion of 5 amino acids had only a relatively small effect, the deletion of 10 amino acids resulted in almost complete activity loss (Figs. 3 and 4). Although neither the tertiary nor secondary structures of SCPS are precisely known, there are reasons to believe that a region located within -30 amino-terminal residues harbors an amphipatic a-helix. Results obtained from CD spectral analyses showed that secondary structure of SCP2 is composed of -40% a-helix (see Fig. 7), and secondary structure prediction using different methods revealed especially high relative helix probabilities for residues 4-32. Therefore, it is reasonable to believe that part of the a-helical structure of SCPB is located close to its amino terminus which was affected by deletion mutagenesis (Fig. 5). The amphipathicity of the putative helix is indicated in Fig. 5A. To test the hypothesis of whether substitution mutagenesis within this proposed structure would also influence the activity of SCP2, we replaced Leu20, which is located adjacent to Ala9 in the Edmundson wheel projection shown in Fig. 5A, with Glu (G1u2O). This substitution, introducing a charged side chain into the hydrophobic surface of the putative amphipatic helix  ...,... l o . . . . , . .. 1 0 ....,... 3 0 . ...,... 40  (Fig. 5A ), led to almost complete inactivation of all three measured activities (Fig 6). However, the exact structural effect of the G1uZ0 substitution cannot be predicted. The CD spectrum obtained for the GW0 variant differed considerably from the characteristic SCP2 wt spectrum (Fig. 7), and secondary structure prediction using the ALB method (53) projected major alterations (Fig. 6B). Therefore, it is possible that a major structural rearrangement within the affected region is responsible for the inactivating effect of the G1u2" substitution.

Q S S S A S W~~I Q 5 B Q~I G G I ? A T K V K
The region from Met"'" up to the carboxyl terminus seems to be of minor importance because carboxyl-terminal deletion of 18 amino acids resulted in high residual activities. In contrast, further deletion of an additional 5 amino acids resulted in almost complete activity loss (Figs. 3 and 4). Therefore, residues between positions 100 and 104 seemed to be essential for SCPS activity. To assess further the specific role of the region in the vicinity of residue 104 in the activity of SCP2, residues a t positions 104, 105, and 106 were replaced by amino acids that were not expected to alter the overall structure of the protein as dramatically as could be expected for the deletion of multiple residues from the carboxyl terminus. If hydrophobicity and the size of the side chain a t Metlo" were increased by replacing Leu (Leu"")), sterol carrier and cholesterol and PC transfer activities of the variant were hardly affected (Fig. 6). Conversely, replacingAsnIo4 with Ile (Ilelo4), which introduced a hydrophobic side chain of similar size at the neighboring residue, led to almost complete loss of sterol carrier activity and to a drastic reduction of PC and cholesterol transfer activities. A similar inactivation of the two latter activities was also obtained if Asnlo4 was replaced by Asp (Asplo"). However, this substitution, which introduced a negative charge without drastically altering the overall size of the side chain, hardly affected the sterol carrier activity. All measured activities could be almost completely abolished, if in addition to AsnIo4, Glylo6 was replaced by Asp AS^'^^''^^). These results show that the structural arrangement of the region in the vicinity of Asnlo4 plays an important role modulating in vitro sterol carrier and cholesterol and PC transfer activities of SCP2.
The function ~f A s n '~~ could be to stabilize the tertiary structure of SCPS via hydrogen bonding. However, this seems unlikely because Asp"" substitution resulted in relatively high residual activities, especially with regard to sterol carrier activity (Fig. 6). More likely is a specific local effect of the inactivating IleIo4 and partly inactivating Asplo' substitutions because CD spectra obtained for wt SCP2 and the Asp"" or IleIo4 variants were remarkably similar (Fig. 7), providing &-helical fractions ranging from 38 to 43% evaluated according to Provencher and Gliickner (15). It is not clear, however, why the Asplo4 substitution hardly affected the sterol carrier activity, whereas it drastically reduced cholesterol transfer activity (Fig. 6). We hope that elucidation of the exact tertiary structure of SCp2 and its interaction with different transport substrates, currently under way, will help explain this apparent discrepancy.
In a previous study (27), it was proposed that Cys7' is important for the lipid transfer activity of SCP2. It was suggested that dimerization via the thiol group may be an essential step in SCP2-mediated lipid transport. To test this hypothesis, we also replaced Cys7' with Val. Although we found that the Val71 variant of SCP2 had a slightly lower sterol carrier activity than wt recombinant SCP2, residual activity of about 90% shows that thiol function by itself is not essential (Fig. 6). This conclusion was also supported by identical results obtained for a Cys7' to Ser substitution (data not shown). On the other hand, it may be possible that the overall side chain structure around Cys71 plays a role for SCP2 activity. Replacing Asp70 with Asn caused striking almost complete inactivation (Fig. 6). However, the latter result could also be because of a major structural rearrangement caused, for example, by removing a structurally important salt bridge in this mutant. The CD spectrum of the variant clearly differed considerably from the characteristic SCp2 wt spectrum (Fig. 71, whereas the secondary structure, according to the prediction shown in Fig. 5B, should not be drastically affected. The conclusion that Cys71 is not crucially involved in the lipid transfer activity of SCPS is supported further by the fact that the amino acid corresponding to Cys71 is replaced by Val in the active lipid transfer protein PXP18, representing the C. tropicalis homologue of SCPS (25).