Structural insights on cholesterol endosynthesis: Binding of squalene and 2,3-oxidosqualene to supernatant protein factor

Journal of Structural Biology xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Journal of Structural Biology journal homepage: www.elsevier.com/locate/yjsbi

Structural insights on cholesterol endosynthesis: Binding of squalene and 2,3-oxidosqualene to supernatant protein factor Monika Christen a, Maria J. Marcaida b, Christos Lamprakis a, Walter Aeschimann a, Jathana Vaithilingam a, Petra Schneider c, Manuel Hilbert d, Gisbert Schneider c, Michele Cascella e,⇑, Achim Stocker a,⇑ a

Department of Chemistry and Biochemistry, University of Bern, Freiestrasse 3, CH-3012 Bern, Switzerland École Polytechnique Fédéral Lausanne, Station 19, CH-1015 Lausanne, Switzerland Institute of Pharmaceutical Sciences, ETH Zürich, Vladimir-Prelog-Weg 4, CH-8093 Zürich, Switzerland d Laboratory of Biomolecular Research (LBR), Paul Scherrer Institut, Villigen-PSI CH-5232, Switzerland e Department of Chemistry and Centre for Theoretical and Computational Chemistry (CTCC), University of Oslo, P.O. Box 1033 Blindern, N-0315 Oslo, Norway b c

a r t i c l e

i n f o

Article history: Received 25 February 2015 Received in revised form 8 May 2015 Accepted 11 May 2015 Available online xxxx Keywords: Cholesterol biosynthesis CRAL-TRIO Free energy perturbation Lipid binding SEC-14 like

a b s t r a c t We present the crystal structures of the SEC14-like domain of supernatant protein factor (SPF) in complex with squalene and 2,3-oxidosqualene. The structures were resolved at 1.75 Å (complex with squalene) and 1.6 Å resolution (complex with 2,3-oxidosqualene), leading in both cases to clear images of the protein/substrate interactions. Ligand binding is facilitated by removal of the Golgi-dynamics (GOLD) C-terminal domain of SPF, which, as shown in previous structures of the apo-protein, blocked the opening of the binding pocket to the exterior. Both substrates bind into a large hydrophobic cavity, typical of such lipid-transporter family. Our structures report no specific recognition mode for the epoxide group. In fact, for both molecules, ligand affinity is dominated by hydrophobic interactions, and independent investigations by computational models or differential scanning micro-calorimetry reveal similar binding affinities for both ligands. Our findings elucidate the molecular bases of the role of SPF in sterol endo-synthesis, supporting the original hypothesis that SPF is a facilitator of substrate flow within the sterol synthetic pathway. Moreover, our results suggest that the GOLD domain acts as a regulator, as its conformational displacement must occur to favor ligand binding and release during the different synthetic steps. Ó 2015 Elsevier Inc. All rights reserved.

1. Introduction Cholesterol and related sterol analogs are major lipid components of eukaryotic cell membranes and essential to evolution of higher organisms Bloch (1983). Cholesterol can be taken up from the diet or alternatively be endogenously synthesized (Ikonen, 2008). Under western diet conditions, roughly two thirds of the total body cholesterol are of endogenous origin (Dietschy and Wilson, 1970). Deregulation of cholesterol levels in the human body has been directly associated to vascular diseases (Ikonen,

Abbreviations: CD, circular dichroism; CRALBP, cellular retinaldehyde-binding protein; ER, endoplasmatic reticulum; FEP, free-energy perturbation; GOLD, Golgi dynamics; MD, molecular dynamics simulations; microDSC, differential scanning micro-calorimetry; NF1, Neurofibromatosis-inducing factor 1; RMSD, root-mean-square deviation; SEM, standard error of the mean; SPF, supernatant protein factor; a-TTP, a-tocopherol transfer protein; vdW, van-der-Waals. ⇑ Corresponding authors. E-mail addresses: [email protected] (M. Cascella), [email protected] (A. Stocker).

2008), which, in turn, are enlisted among the first causes of mortality in developed countries. It is thus of major importance to understand the molecular mechanisms that regulate the cholesterol synthesis (Maxfield and Tabas, 2005). The late stages of the committed cholesterol synthesis involve the formation of squalene, its oxygen-dependent epoxidation to 2,3-oxidosqualene, and the subsequent cyclization of the epoxide leading to the sterol backbone (Chugh et al., 2003). The epoxidation reaction by the microsomal squalene monooxygenase has been extensively studied for more than twenty years by Konrad Bloch and colleagues (Yamamoto and Bloch, 1970). Initially, rat liver microsomes were reported to require supernatant fraction (S105) for monooxygenase activation (Tchen and Bloch, 1957). Subsequent studies identified and characterized specific activators of the reaction, NADPH-cytochrome-c reductase, NADPH, FAD, phospahtidylserine, phosphatidylglycerol and a heat-labile protein (Tai and Bloch, 1972; Yamamoto and Bloch, 1970). The heat-labile protein was later purified from rat cytosolic fraction and termed ‘‘supernatant protein factor’’ (Ferguson and Bloch, 1977). Purified

http://dx.doi.org/10.1016/j.jsb.2015.05.001 1047-8477/Ó 2015 Elsevier Inc. All rights reserved.

Please cite this article in press as: Christen, M., et al. Structural insights on cholesterol endosynthesis: Binding of squalene and 2,3-oxidosqualene to supernatant protein factor. J. Struct. Biol. (2015), http://dx.doi.org/10.1016/j.jsb.2015.05.001

2

M. Christen et al. / Journal of Structural Biology xxx (2015) xxx–xxx

rat SPF actively promoted squalene transfer across membranes in vitro and it was proposed to be directly involved in substrate transfer towards squalene monooxygenase (Friedlander et al., 1980). In 1999 we identified a 46 kDa protein by incubating cytosol of bovine liver with radioactively labeled a-tocopherol (Stocker et al., 1999). The human homolog of this protein was cloned and initially termed tocopherol associated protein (TAP) due to its ability to bind a-tocopherol (Zimmer et al., 2000). One year later Arai and co-workers showed that TAP was identical to SPF (Shibata et al., 2001). They also reported the primary structure of SPF, its in vitro squalene transfer activity and its in vivo stimulatory role for sterol production in SPF-transfected hepatoma cells. By use of a SPF-deficient mouse model, the same group confirmed that SPF plays an in vivo role in hepatic cholesterol synthesis under fasting conditions (Shibata et al., 2006). Recent investigations on human SPF in rat hepatoma cells also show that phosphorylation may play a role on its activation (Mokashi and Porter, 2005). We reported recently the crystal structure of SPF revealing its two-domain architecture consisting of a SEC14-like lipid-binding core-domain and a C-terminal jelly-roll barrel with putative functions in Golgi dynamics and in secretion (Anantharaman and Aravind, 2002; Stocker et al., 2002). Surprisingly, no evidence for direct squalene binding by SPF has been reported to date (Chin and Bloch, 1985; Shibata et al., 2001; Stocker and Baumann, 2003; Stocker et al., 1999). Thus, the in vivo ligand specificity of SPF and its physiological role are still unresolved. Here we present high-resolution X-ray structures of the SEC14-like domain of SPF in complex with its proposed natural ligands squalene and 2,3-oxidosqualene. The removal of the C-terminal jelly-roll barrel of SPF has proven favorable for ligand-binding and allowed us to experimentally assess the thermal stability by circular dichroism (CD) spectroscopy and measure the thermodynamic parameters of binding by differential scanning micro-calorimetry. Ligand binding was also studied by molecular dynamics simulations (MD) and free-energy perturbation calculations, following protocols successfully applied to investigate substrate selectivity by a-tocopherol transfer protein (a-TTP) (Helbling et al., 2012, 2014). Our calculations show that SPF has no significant preferentiality for either squalene or 2,3-oxidosqualene, in agreement with experimental data. Overall, our results support the hypothesis originally proposed by Bloch and co-workers that SPF facilitates cholesterol synthesis by acting as dual substrate carrier that demonstrate the ability of SPF to stimulate squalene monooxygenase. However, the binding promiscuity of SPF for other lipids evidences the need for further studies to fully understand its function. 2. Materials and methods 2.1. Expression and purification SPFs (residues Met1 to Lys275) was cloned into the pET28a vector (Novagen) by site directed mutagenesis using the full length SPF sequence as template. The oligonucleotides (Microsynth) used for

PCR

were

50 -GGGAATTCCATATGAGCGGCAGAGTC-30

and

5 -GTGTTCCTCGAGCTATTTCACCTGGTC-30 with the NdeI and Xhol sites underlined and the introduced stop codon in bold. The construct was transformed into Escherichia coli BL21(DE3)Lemo (NEB). Protein expression was carried out following standard auto-induction protocols using ZYP-5052 media supplemented with antibiotics in a HT Minifors Fermenter (Studier, 2005). An additional lactose feed was introduced and protein was expressed at 22 °C. The cells were harvested at 5000 rpm for 25 min at 4° C and resuspended in lysis buffer (20 mM imidazole, 300 mM NaCl,

20 mM Tris–HCl, 1 mM PMSF, pH 7.4). Disruption of the cells was performed by sonication cooled on ice, after which the debris were removed by centrifugation at 50,000 rpm for 30 min at 4 °C. SPFs was purified by affinity chromatography using a His-TrapFF column from GE Healthcare following manufacturer’s instructions. After loading, the His-TrapFF column was washed with lysis buffer, followed by a second wash with 6% elution buffer (500 mM imidazole, 300 mM NaCl, 20 mM Tris–HCl, pH 7.4). Protein was eluted with 100% elution buffer. The eluate was exchanged into final buffer (40 mM NaCl, 20 mM Tris–HCl, pH 7.4) using a HiPrep 26/10 desalting column from GE Healthcare. The His-tag was cleaved with thrombin protease (5 units per mg of SPFs) at 4 °C overnight. The tag and minor impurities were removed by size exclusion chromatography using a HiLoad 16/60 Superdex™ 75 column (GE Healthcare). SPFs was concentrated using Vivaspin-15 concentrators (Sartorius Stedim Biotech). Protein concentration was determined at 280 nm by UV absorbance (e280 = 45,380 M1 cm1) using a Nanodrop2000 from Thermo Scientific. SPFs was flash frozen in liquid nitrogen and stored at 18 °C. SPF full length was expressed and purified following the same procedures. 2.2. Preparation of protein–ligand complexes The natural ligands squalene and 2,3-oxidosqualene were purchased from Sigma (P98% and P92% purity, respectively). For solubilization, 60 mg of sodium cholate were overlaid with 1.5 mg of corresponding ligand and centrifuged at room temperature for 5 min at 13,600 rpm yielding an oil in detergent matrix. 250 ll of final buffer were added and the resulting suspension containing 557.6 mM sodium cholate and 14.6 mM ligand was sonicated in a water bath, until it became transparent. The solubilized ligand was mixed with purified SPFs yielding 27.9 mM sodium cholate, 1 mg/ml protein with a 25-fold molar excess of ligand over protein. This mixture was dialysed 3 times for 5 h against final buffer at 4 °C. Eventually forming precipitates were removed by centrifugation at 5000 rpm and the remaining ligand–protein complex was concentrated and unbound ligand removed by size exclusion chromatography. Complex formation was checked using GC–MS (squalene) and nano-ESI-MS (2,3-oxidosqualene). In both cases 0.2 mg of complex were lyophilised and extracted with methanol. As negative control, apo SPFs was analyzed under equal conditions. GC-chromatograms of squalene were carried out on an Agilent Technologies, 7820 A GC system on a Supelco 28045-4 column (15 m  250 lm  25 lm). A temperature ramp of 180–250 °C (5 °C/min) and a flow rate of 1.5 ml/min were applied. The detection was carried out using a FID detector at 310 °C. Data acquisition was done at a data rate of 50 Hz/0.004 min. A peak at 13.6 min could be readily seen in squalene controls as well as in the extracted samples (Fig. S2A–C). These were submitted to MS (Finnigan Trace MS, Thermo Quest) and the obtained fragments were compared with the SDBS-Mass database (Anantharaman and Aravind, 2002), confirming the presence of squalene in the complex (Fig. S2D). As shown in Fig. S2E–G, nanoESI (Thermo Scientific LTQ Orbitrap XL) in positive ion mode was used to detect 2,3-oxidosqualene. Resulting spectra showed a mass peak at m/z = 449.38 indicating the singly charged sodium adduct of 2,3-oxidosqualene.

0

2.3. Crystallization Protein–ligand complexes were concentrated to 12 mg/ml (squalene) and 17 mg/ml (2,3-oxidosqualene) using the Vivaspin-15 concentrators (Sartorius Stedim Biotech). The crystals were obtained by hanging drop vapor diffusion method at 18 °C, whereby drops were set up by mixing 1 ll of sample solution with 1 ll of reservoir solution (squalene: 0.2 M Lithium sulfate, 0.1 M

Please cite this article in press as: Christen, M., et al. Structural insights on cholesterol endosynthesis: Binding of squalene and 2,3-oxidosqualene to supernatant protein factor. J. Struct. Biol. (2015), http://dx.doi.org/10.1016/j.jsb.2015.05.001

M. Christen et al. / Journal of Structural Biology xxx (2015) xxx–xxx

Bis-Tris, 25% PEG3350, pH 6.5/2,3-oxidosqualene: 1 M Ammonium sulfate, 0.1 M Bis-Tris, 1% PEG3350, pH 5.5) and equilibrated against 500 ll of reservoir solution. Typically, after 3–5 days plate-like crystals grew to maximum size (1  0.3  0.05 mm), and were flash-frozen in liquid nitrogen under cryo-protection (reservoir solutions with 25% glycerol). Data were collected on the X06DA beamline at the Swiss Light Source, Paul Scherrer Institut, Villigen, Switzerland.

3

ligands was checked using simulated annealing omit maps in Phenix. Several rounds of refinement (Phenix (Adams et al., 2010) and Refmac5 (Winn et al., 2011)) and model building (Coot (Emsley et al., 2010)) were then carried out until convergence was reached. Residues GSH at the N-terminus and K275 at the C-terminus could not be modeled. In the final models, no Ramachandran outliers are present. Figures were made with Pymol (DeLano, 2002), CHIMERA (Pettersen et al., 2004) and LIGPLOT (Wallace et al., 1995).

2.4. Structure solution and refinement Data collection was carried out at the Swiss Light Source PXIII beamline (PSI, Villigen). The data were integrated and scaled with MOSFLM/SCALA (Evans, 2006; Leslie and Powell, 2007) or XDS (Kabsch, 2010) (Crystallographic data statistics are shown in Table 1). The structures were solved by molecular replacement using Phaser in CCP4 (Winn et al., 2011). The model used was the truncated structure of SPF full length (PDB 1O6U) with an empty cavity. In both structures, two molecules per asymmetric unit are present (related by 2-fold NCS). A first round of rigid body refinement was performed, followed by simulated annealing (Adams et al., 2010) and initial electron density maps showed the presence of the ligands. The ligands were fitted using LigandFit in Phenix (Adams et al., 2010). The conformation of the Table 1 Crystallographic data collection and refinement statistics.

X-ray source & detector Wavelength (Å) Space group Cell (Å) Resolution range (Å)* Number of observations* Unique reflections* Completeness (%)* Rmerg (%)* Rpim (%)* Mean I/r(I)* Multiplicity*

SPFs-2,3oxidosqualene

SLS PXIII Pilatus 2M 0.999 P21 71.27, 73.54, 94.23 90.00, 96.88, 90.00 30.00–1.75 (1.84–1.75) 304,942 (46,480) 97,261 (14,143) 99.7 (99.8) 5.2 (31.7) 3.5 (20.5) 12.7 (3.6) 3.1(3.3)

SLS PXIII Pilatus 2M 0.999 P21 71.88, 73.80, 94.45 90.00, 96.29, 90.00 29.89–1.60 (1.69–1.60) 400,409 (54,851) 128,890 (18,675) 99.7 (99.3) 5.4 (28.8) 3.7 (19.6) 11.1 (3.2) 3.1(2.9)

Refinement statistics (Phenix Refine) Resolution (Å) 29.88–1.75 R (working set) (%) 14.82 Rfree (%) 17.45 Number of reflections 179,461

29.89–1.60 15.44 17.85 128,780

RMSD from ideal values Bond lengths (Å) Bond angles (°)

0.010 1.706

0.010 1.336

4504 2  squalene (101 atoms) 2  IPA, 2  Cl, 6  SO2 4 853

4565 2  2,3oxidosqualene (62 atoms) 9  Cl, 7  SO2 4 1062

24.2 33.7

20.9 30.7

36.9

34.8

98.9 0

99.26 0

1.95 (99%) 0.96 (100%)

1.92 (99%) 0.96 (100%)

Number of atoms Protein Non-protein

Water B factor (Å2) Protein Squalene/2,3oxidosqualene Water MolProbity statistics Ramachandran favored (%) Ramachandran disallowed (%) Clash score (percentile) MolProbity score (percentile) *

SPFs-squalene

Values in parentheses correspond to the high-resolution shell.

2.5. CD measurements A Jasco J-175 Spectropolarimeter with a Peltier PFD-350S temperature controller was used to measure CD spectra and temperature-depended protein unfolding. A 10 mm path length cell (with 500 ll sample) was used and the protein concentration was 2 lM in final buffer. The response was set to 1 s with a bandwidth of 5 nm. Following the results from the CD spectra, the wavelength was adjusted to 222 nm for temperature-dependent protein unfolding experiments. The temperature was increased at a rate of 2°K min1 from 20 °C to 80 °C and measurements were taken in increments of 0.5°K (Fig. 6). The melting temperatures were calculated from the 1st derivative of the unfolding curves. All measurements were performed in triplets, leading to an average melting temperature and corresponding standard deviation (Fig. 6).

2.6. Computational methods 2.6.1. System Setup The new crystal structures of SPFs in complex with squalene and 2,3-oxidosqualene and well as full length SPF (PDB 1O6U) (Stocker et al., 2002) were used as starting points. Hydrogen atoms were added using the Propka (Rostkowski et al., 2011) software at pH 7.0. The protein was parameterized within the AMBER FF99SB (Hornak et al., 2006) force field while for the ligands the General Amber Force Field (GAFF) (Wang et al., 2004) was used. Moreover, ligands’ charges were assigned following the RESP procedure (Cieplak and Kollman, 1993). The protein systems were solvated by TIP3P water molecules (Jorgensen et al., 1983), and three chlorine anion atoms were added to neutralize the system using Aqvist parameters (Aqvist, 1990). The initial boxes for SPFs simulations had a dimension of 100  110  100 Å. For the full-length SPF squalene complex, the ligand was placed by structural alignment of the N-terminal SEC-14 like domain to the SPFs structure.

2.6.2. Molecular dynamics Molecular dynamics simulations were performed with the NAMD package (http://www.ks.uiuc.edu/Research/namd). The systems described above were minimized during 2000 steps of steepest descent algorithm, and were slowly heated from 0 K to reach 300 K temperature within a 120 ps run of NVT ensemble. Another 1000 steps course of steepest descent was applied to minimize again the structures. The systems then were equilibrated during a 100 ns run in the NPT ensemble at a temperature of 300 K and pressure 1 bar controlled by modified Nose–Hoover (Hoover, 1985; Martyna et al., 1992; Nose, 1984) and Langevin pistons (Feller et al., 1995) thermo-barostats. The Particle Mesh Ewald method (Essmann et al., 1995) was used to compute the full system periodic electrostatics. Moreover, all bonds between hydrogen and heavy atoms were constrained to their equilibrium length using the SHAKE algorithm (Ryckaert et al., 1977). The time step was set to 2 fs.

Please cite this article in press as: Christen, M., et al. Structural insights on cholesterol endosynthesis: Binding of squalene and 2,3-oxidosqualene to supernatant protein factor. J. Struct. Biol. (2015), http://dx.doi.org/10.1016/j.jsb.2015.05.001

4

M. Christen et al. / Journal of Structural Biology xxx (2015) xxx–xxx

2.6.3. Free energy perturbation calculations (FEP) The relative binding affinity between the squalene, and the 2,3-oxidosqualene in different head or tail binding conformations were estimated following standard FEP protocols (Kollman, 1993). FEP transforming squalene into 2,3-oxidosqualene either in the tail or in the head bound conformations were performed. 7 k-points were used for each transformation. Backward transformations were repeated to take into account hysteresis in the estimation of the free-energy difference. For each of them l-point, the system was relaxed following standard minimization procedures as described above, and then let run for a production simulation of 20 ns. The FEP simulations and analysis was performed with the GROMACS package (Hess et al., 2008). 2.7. Differential scanning calorimetry Differential scanning calorimetry (DSC) was performed on a TA Instruments Nano DSC running Nano DSCRun Software v4.2.4 data acquisition software. Each DSC experiment required 300 ll of protein sample as well as the same volume of reference buffer (20 mM TRIS–HCl, 40 mM NaCl, pH 7.4). Each sample was measured in duplicates and protein concentration of the samples was determined in a range between 0.3846 and 0.5713 mg/ml by amino acid analysis prior to scanning. Samples were centrifuged at 13,200 rpm for 10 min at 4 °C and degassed for five minutes prior to DSC experiments. Reference buffer was used to generate a baseline and subtracted from the sample DSC thermograms. Data were collected at 3 bar pressure between 20 and 80 °C in temperature increments of 2 K min1 with 10 min pre-equilibration prior to scanning. Raw data processing and visualization was performed with NanoAnalyze Software v3.1.2. Protein stability curves, thus the free energy of the fold (DGu), were calculated using the modified Gibbs–Helmholtz equation and the thermodynamic (Tm, DHu,m and DCpu) values from the DSC experiments (Bruylants et al., 2005). 3. Results and discussion Despite a huge body of evidence for SPF in mediating substrate transfer in the committed cholesterol synthesis, its physical interaction with squalene or 2,3-oxidosqualene has not been described yet (Chin and Bloch, 1985; Shibata et al., 2001; Stocker and Baumann, 2003; Stocker et al., 1999). Careful inspection of the known crystal structure of SPF has been crucial to resolve this issue by identifying a region of the protein that impairs access of the substrate to the ligand-binding moiety (Stocker et al., 2002). A similar case of self-inhibition within a SEC14-like fold has been previously described in the case of Neurofibromatosis-inducing factor 1 (NF1) (D’angelo et al., 2006). The two-domain fold of SPF includes an N-terminal SEC14-like lipid-binding domain and a C-terminal jelly-roll motif (Fig. 1). The SEC14-like moiety of SPF is composed of a three-helix bundle and a three-layer a-b-a-sandwich. This is a conserved domain (Pfam 00650) with functionality in binding small lipophilic molecules as seen in other single-domain SEC14-like family members including the prototypic yeast phosphatidylinositol/phosphatidyl choline transfer protein (SEC14) (Sha et al., 1998), a-TTP (Meier et al., 2003) and cellular retinaldehyde-binding protein (CRALBP) (Bolze et al., 2014; Cascella et al., 2013; He et al., 2009). The additional C-terminal jelly-roll barrel in SPF has recently been referred to as the Golgi dynamics (GOLD)-domain (Anantharaman and Aravind, 2002). The barrel is composed of two four-stranded antiparallel b-sheets and a short C-terminal helical extension. As shown in Fig. 1, the C-terminus of SPF spans the SEC14-like domain

α9

α12

α10

Fig.1. Structural constraints for the opening of the mobile gate in SPF full length. The GOLD-domain (yellow) of SPF clamps the SEC14-like domain (light gray) from above. The N-terminal 3-helix bundle is shown in light blue. The insert zooms into the mobile gates of SPF (orange) and the superimposed aTTP (blue) in closed and open conformation, respectively. The structural clash of the open gate and helix a12 is shown.

from the top in a clamp like conformation. Structure based sequence alignments show highest structural similarities between the SEC14-like domains of SPF and of a-TTP (alignment of primary and secondary structures shown in Fig. S1). a-TTP, as well as SEC14, has been reported to exist in two conformations that differ mostly in the hinge movement of a flexible region referred to as mobile gate, that permits access through the entrance region to the hydrophobic cavity (Meier et al., 2003; Ryan et al., 2007; Schaaf et al., 2008). In a-TTP, the mobile gate (residues 199–220) adopts an open (PDB 1OIZ) (Meier et al., 2003) or a closed state (PDB 1OIP, PDB 1R5L) (Meier et al., 2003; Min et al., 2003) depending on the presence of the ligand. A superposition of SPF to both conformations of a-TTP shows that the closed state of the mobile gate of SPF (residues 194–215, helices a9, a10) is locked in place by its GOLD domain (Fig. 1). Therefore, we hinted that the propensity to bind ligands of very high hydrophobicity such as squalene (Hauss et al., 2002) to SPF in vitro is low because of structural constraints. Thus, we designed the construct SPF-short (SPFs), comprising residues 1–275, in which the entire GOLD-domain including the C-terminal helix a12 is removed. In this truncated construct, steric hindrance acting on the mobile gate is released, and thus, substrates may have an easier access to the cavity. In order to test the effect of deleting the GOLD domain, we performed molecular dynamics simulations (MD) on SPFs compared to SPF full length (MD comparison of mobile areas shown in (Fig. 2). Despite the overall fold being well conserved, our simulations report a significant increase in the mobility of two segments of SPFs between residues 196 and 203 (helix a9), and between residues 219 and 224, located at the loop connecting b5 to a11, both segments being significantly more mobile, with a relative increase of the RMSF of the corresponding amino acids between 70% and 100% (Fig. 2). These regions are in direct contact with the C-terminal helix a12, in SPF full length (Fig. 1). Therefore, our simulations confirm that the opening of the cavity is intrinsically facilitated by the displacement of the GOLD domain from the SEC14-like core.

Please cite this article in press as: Christen, M., et al. Structural insights on cholesterol endosynthesis: Binding of squalene and 2,3-oxidosqualene to supernatant protein factor. J. Struct. Biol. (2015), http://dx.doi.org/10.1016/j.jsb.2015.05.001

M. Christen et al. / Journal of Structural Biology xxx (2015) xxx–xxx

Fig.2. Molecular dynamics simulations of mobile areas on SPFs compared to SPF full length. The root mean square fluctuation (RMSF) values for the protein Ca atoms are shown for the SPF-squalene full length (black) and for the SPFs-squalene (red) complexes. Differences in the mobility can be seen in the mobile gate (orange) and in the last C-terminal short helix of the SEC14-like domain, together with the adjacent areas comprising residues 76–82 and 111–123 (blue). The corresponding areas are highlighted in the structure of SPF full length that is shown in cartoon representation. On the contrary and somewhat surprising, the helical portion between residues 160 and 165 (helices g2/a8) does not show any change in mobility despite its close topological proximity to the GOLD domain. In addition, no differences were observed for the very well conserved residues KPFL that are part of a short 310 helical motif that links helices a9 and a10. These form a compact structural signature of the mobile gate region in SEC14-like proteins that is not likely to change conformation. See also Fig. S1.

3.1. Crystal structures of SPFs bound to squalene and 2,3oxidosqualene The complexes between SPF/SPFs and squalene or 2,3-oxidosqualene were prepared as described in the methods section. SPFs bound to squalene and 2,3-oxidosqualene yielded crystals that diffracted to 1.75 and 1.60 Å resolution and the structures were solved by molecular replacement using the model of SPF full length (PDB ID: 1OLM). The crystallographic and refinement statistics are shown in Table 1. The chemical identity of the ligands bound to SPFs was proven by GC–MS and nano-ESI mass spectrometry for squalene and 2,3-oxidosqualene, respectively (Fig. S2). The presented model of both crystal structures includes residues 1–274 (Fig. 3A and B). The superposition of the SPFs complexes with SPF full length revealed minimal deviations with RMSD values over all atoms of 0.9 Å and of 0.8 Å for the structures with squalene and 2,3-oxidosqualene, respectively. Both structures are in the closed conformation even though the C-terminal helix of the GOLD-domain is not present. The low RMSD values of SPFs compared to SPF full length indicate that this truncated construct is a reliable system for in vitro ligand binding studies. Electron density for the ligand was already observed in the first molecular replacement maps within the horseshoe shaped ligand binding cavity in both complexes (Fig. 3C). In the refined structure of SPFs bound to squalene, the density map of the ligand is well defined close to the entrance of the binding pocket, but suggested the presence of two different conformations towards the inner core of the cavity. In the case of 2,3-oxidosqualene, the crystal structure revealed strictly only one binding mode, with the epoxide group leaning towards the entry of the cavity. Again, the isoprene tail pointing to the inner core displayed weaker density, implicating higher flexibility for the tail of the molecule. Overall, the interactions between ligand and protein within the binding cavity were found to be exclusively hydrophobic and

5

highly similar in both complexes (Fig. 4). This is confirmed by MD-calculations reporting an average interaction energy of 70 kcal mol1 for both substrates (for single contributions of the aa forming the binding cavity, see Table S1). Specifically, the entrance region including the mobile gate provided the highest number of hydrophobic contacts that locked the corresponding ligand in place. Accordingly, the inner face of the mobile gate helix a9 has been reported to be engaged in vdW interactions with bound ligands in all known crystal structures of SEC14-like family members so far (He et al., 2009; Meier et al., 2003; Schaaf et al., 2008; Welti et al., 2007). This explains the stronger electron density in this region of SPFs for squalene and 2,3-oxidosqualene (Fig. 3C). In previous MD studies on CRALBP (Helbling et al., 2013), we reported significant differences in the binding properties of the ligands at the temperature of the crystal and at room temperature. Therefore, we repeated independent MD runs in SPFs both at 300 K and at 100 K, which corresponds to roughly the temperature of the crystal. All MD runs report a stable fold for the SEC14-like domain over more than 100 ns, with RMSDs for the Ca that in all cases oscillate around 1.5 Å (Fig. S3). At 100 K the SPFs structure is rather rigid. At this temperature, the RMSD of the carbon atoms of both ligands deviate from the crystal structure positions on average by 0.4 Å, consistently with what was found for the Ca atoms of the protein (RMSD = 0.5 Å). At 300 K, the absence of any specific directional contacts allows for exploration of multiple ligand conformations. Cluster analysis on the conformations observed during MD revealed that squalene adopts mostly one conformation in SPFs (cluster population of 92%) while a second conformation is seen for 5% of the simulation time (Fig. 3D). MD simulations of SPFs bound to 2,3-oxisqualene identified again the presence of a dominating conformation (83% relative abundance) with two other thermally accessible geometries populated during 12% and 3% of the simulation time (Fig. 3E). Our cluster analysis evidences the presence of multiple conformations in the region of the tail isoprene for 2,3-oxidosqualene. It can leave the position observed in the crystals, localized between helices a7 and a8, to form a contact with Leu111 and Leu84. This conformation and the crystallographic one dynamically exchange within tens of ns, implying that they are both accessible at room temperature. This is consistent with the less defined electron density observed in the X-ray scattering experiment in that area of the ligand. 3.2. Molecular background of dual substrate specificity in SPFs Overall, three common ligand-interacting regions of SPFs could be identified in both complexes. The first region is represented by residues within the back face of the cavity belonging to helices a6 and a7 (Leu106, Ala108, Leu111, Leu120 and Lys124 shown in magenta, Fig. 4). The second region constitutes the central walls of the cavity including strand b3 and helix a8 and the connecting loop (Tyr153, Cys155, Leu158, His162, Ala167, Ala170, Tyr171, Leu175 and Leu189 shown in blue, Fig. 4). The third region defines the front face of the cavity through residues within helix a9 of the mobile gate (Phe198, Ala201, Tyr202 and Ile205 shown in orange, Fig. 4). Of 20 residues contacting the ligands 18 are identical between squalene and 2,3-oxidosqualene. Two vdW interactions different in squalene were identified in helix a10 (Thr213 and Ile217 shown in yellow, Fig. 4B) and two other for 2,3-oxidosqualene in helix a8 and strand b3 (Phe174 and Ile151 respectively, both highlighted in cyan, Fig. 4C). MD-calculations show that among the large number of residues composing the binding cavity (Table S1), the two H-bonded Tyr153 and Try171 in the middle of the binding cavity contribute to most of the

Please cite this article in press as: Christen, M., et al. Structural insights on cholesterol endosynthesis: Binding of squalene and 2,3-oxidosqualene to supernatant protein factor. J. Struct. Biol. (2015), http://dx.doi.org/10.1016/j.jsb.2015.05.001

6

M. Christen et al. / Journal of Structural Biology xxx (2015) xxx–xxx

A

B

C

D

E

Fig.3. Structures of SPFs bound to squalene and 2,3-oxidosqualene. (A) Structure of SPFs in complex with squalene (purple). The ligand was modeled in two alternative conformations. (B) Structure SPFs in complex with 2,3-oxidosqualene (green). In both cases, the mobile gate and helix a8 have been rendered semi-transparent to better visualize the ligands. (C) Electron density (2Fo  Fc coefficients) of the ligands shown at r = 1. (D and E) Cluster analysis on the conformations observed during MD for squalene and 2,3-oxidosqualene, respectively. The thickness of the model is proportional to the abundance of that particular conformation during the simulation time.

A

Ile151

C 6

Leu189

Ala 108 His 162

Leu 111

Leu 158

Ala 167

Phe174 Ile205

Phe 198

Leu 106

Ala201

Ala 201 Ala 170 Tyr 171

Tyr 202 Cys 155

3

Leu 120

Tyr153

Ile 205

9

Tyr 153 Phe 174 Leu 175

Tyr202

Leu175 Tyr171

Lys 124 Ile 217 Ile 151

7

8

Leu 189

4

10

Leu158 Phe198 Lys124

Thr 213

Cys155 Ala170

His162 Leu189

B

Thr213

Ala108

Ile205 Ala201

Ala167

Leu106

Ile217

Tyr153

Leu111 Tyr202

Leu175

Leu120

Fig.4. The ligand-binding cavity. (A) Close up view of the ligand-binding pocket: back face (magenta), central cavity (blue) and front face (orange). Residues involved in ligand specific interactions are shown in cyan (2,3-oxidosqualene) and in yellow (squalene). (B and C) Two-dimensional plot of the hydrophobic interactions (2.9–3.9 Å) between SPFs and squalene and 2,3-oxidosqualene, respectively. Figures made with LIGPLOT (Wallace et al., 1995). See also Table S1.

Please cite this article in press as: Christen, M., et al. Structural insights on cholesterol endosynthesis: Binding of squalene and 2,3-oxidosqualene to supernatant protein factor. J. Struct. Biol. (2015), http://dx.doi.org/10.1016/j.jsb.2015.05.001

M. Christen et al. / Journal of Structural Biology xxx (2015) xxx–xxx

interaction energy (4.8 and 8.5 kcal mol1, respectively). These two residues are interacting with each other through either a direct hydrogen bond or through a hydrogen bond mediated by water molecules that cross the cavity. This contact actually bridges the beta sheet b3 with the helix a8 (Fig. 4A). Due to the hydrophobic nature of all interactions no selection criteria specific for one or the other ligand could be identified. In addition, the fact that the shape of the cavity remains constant upon ligand-binding implies that in order to fit, any ligand has to adopt a complementary shape, as previously suggested by Arai and co-workers (Shibata et al., 2001). Despite the fact that crystallography reveals a preferred orientation for 2,3-oxidosqualene, the absence of specific hydrogen-bonding contacts stabilizing the epoxide may suggest the possibility of an alternative binding mode characterized by head-to-tail swapping. Free-energy perturbation (FEP) estimates predict a relative binding affinity between the crystallographic or the alternate position of DDG = 0.96 ± 1.7 kcal mol1. This means that, within the accuracy of the method, the two binding geometries should be practically equally accessible at room temperature. Since according to FEP there is no thermodynamically preferred orientation of the epoxide, we assume that the observed orientation in the crystal is obtained by kinetic selection occurring when working in vitro. Accordingly, the largely hydrophobic patch of the binding cavity exposed to the solvent when the mobile gate is opened, would be expected to facilitate selection of the more hydrophobic tail to enter first. However, we cannot exclude at this stage that the head in front orientation of 2,3-oxidosqualene in SPFs might be also relevant in vivo. In fact, this configuration would intuitively be the preferred one for 2,3-oxidosqualene-transfer in case of direct interaction of SPF with the enzymes involved in the endosynthetic pathway. 3.3. Solvent access to the cavity is thermally controlled Despite its strong hydrophobic character, the binding cavity is partially hydrated when both, squalene or 2,3-oxidosqualene, are loaded. Crystallographic waters detected in our structures suggest the presence of a channel crossing the binding cavity sitting in close proximity to the ligand. Comparison between MD simulations at 100 K and 300 K revealed a substantial opening of this water channel upon temperature increase. This is illustrated in Fig. 5, where water accessible cavities in the protein at different temperatures are shown. This phenomenon was observed in both SPFs and SPF full-length simulations. Water can exchange from the bulk through three gates placed close to the ligands’ ends. The first gate is localized in the neighborhood of the isoprene tail where water can move among the side chains of Leu85, Thr124, Asp102 and

A

B

7

Lys125. Another water passage comprises interstitial space defined by Glu174, Lys118 and Leu122. Finally, water has access to the binding cavity through residues Phe179, Glu185, Leu187 and Glu180. No water molecules were observed at the interface between the ligands and the mobile gate a8 helix, consistently with the strong hydrophobic character of this area. Thus, upon ligand entering or exiting through the mobile gate, water may leave or access the cavity through another path, implicating that the three entrances of the water channel act as a counterbalancing valve. In the fully closed state, the average number of water molecules at a minimal distance of 6 Å around the ligand’s atoms is 11.6 for the squalene and 15.8 for 2,3-oxidosqualene; thus, indicating a larger solvation of the latter (at 300 K). 3.4. SPFs binding affinity for 2,3-oxidosqualene is only slightly stronger than for squalene Data presented here elucidate the molecular means by which SPF directly binds to both cholesterol precursors squalene and 2,3-oxidosqualene. FEP estimates of the relative binding affinity between 2,3-oxidosqualene and squalene in SPFs predict a difference in binding affinity DDG = 2.6 ± 2.1 kcal mol1 in favor of 2,3-oxidosqualene, when virtually extracting the free ligands from hexane. For example, a virtual extraction from water would invert the affinity in favor of the more hydrophobic squalene, with a DDG = 2.2 ± 1.9 kcal mol1. For validating our FEP calculations we complemented our study with CD spectroscopy to monitor temperature-dependent protein unfolding (Greenfield, 2006) and with differential scanning micro-calorimetry (microDSC) (Bruylants et al., 2005) to directly assess thermodynamic parameters of ligand binding. The CD melting temperatures of SPFs-apo (55.0 ± 0.4 °C) and the SPFs-complexes with squalene (61.7 ± 0.5 °C) and with 2,3-oxidosqualene (61.0 ± 0.8 °C) yielded a shift in Tm of roughly 7 °C for both complexes compared to SPSs apo (Fig. 6A). The corresponding transition temperatures (Tm) from the microDSC experiments were in excellent agreement with the results from CD thermal denaturation curves (Fig. 6B). The raw microDSC data shown in Fig. 6C illustrate the heat uptake associated with thermal unfolding of SPFs in the presence and absence of substrate molecules (squalene or 2,3-oxidosqualene). The increase in Tm in the presence of each substrate was consistent with specific ligand binding to the native state of SPFs (Cooper, 1999). The increase in area under each endotherm with higher Tm, and the higher heat capacity baselines after the unfolding transitions, were indications of the significant positive DCp commonly associated with such processes. In addition, the fairly symmetric shape of the endotherms and the absence of either, additional peaks or discontinuities indicated that unfolding of ligand-complexes of SPFs

C

Fig.5. Water channels in SPFs. (A) Space filled representation of the water accessible cavities in the SPFs crystal structure. (B and C) Depiction of how the water channels open up and water reaches the cavity in MD simulations at 100 K and 300 K, respectively.

Please cite this article in press as: Christen, M., et al. Structural insights on cholesterol endosynthesis: Binding of squalene and 2,3-oxidosqualene to supernatant protein factor. J. Struct. Biol. (2015), http://dx.doi.org/10.1016/j.jsb.2015.05.001

8

M. Christen et al. / Journal of Structural Biology xxx (2015) xxx–xxx

A

B

C

Sample apoprotein SPFs SPFs + squalene SPFs + 2,3-oxidosqualene

Tm [K] 329.65±0.12 335.72±0.38 334.61±0.07

Hu,m [kcal mol-1] 222.69±8.98 219.66±7.63 234.54±7.22

Su,m [kcal K-1 mol-1] 0.68±0.03 0.66±0.02 0.70±0.02

Cpu [kcal K-1 mol-1] 2.30±0.09 2.91±0.32 2.30±0.17

Fig.6. Temperature-dependent protein unfolding. (A) Single CD melting curves show the change in the mean residue molar ellipticity at 222 nm as a function of temperature. Triple measurements were performed and average melting temperatures (Tm) were calculated from the 1st derivatives for SPFs apo: 55.0 ± 0.4 °C, SPFs-squalene: 61.7 ± 0.5 °C and SPFs-2,3-oxidosqualene 61.0 ± 0.8 °C. (B) The table highlights thermodynamic parameters (transition temperatures (Tm), enthalpy of unfolding DHu,m and heat capacity of transition DCpu) for the unfolding of SPFs apo and its ligand complexes by microDSC. Data were monitored in duplicates and are denoted as averages ± SEM. (C) Thermograms of single microDSC experiments monitored as a function of temperature for SPFs apo and the corresponding complexes with squalene and 2,3-oxidosqualene, respectively.

proceeds via a reversible 2-state process. A comparison of FEP estimates determined at 300 K (DDG = 2.6 ± 2.1 kcal mol1) with the corresponding difference in binding affinity by microDSC (DDG = 2.36 kcal mol1) was consistently in favor of 2,3-oxidosqualene (Table S2). The microDSC data are in principle suitable to derive contributions of DHb and TDSb to the free energy of binding DGb for squalene and for 2,3-oxidosqualene complexed to SPFs (Bruylants et al., 2005). For this, the enthalpy of binding (DHb) and the entropy of binding (DSb) were determined by subtracting the corresponding enthalpies (DapoHu  DholoHu) and entropies (DapoSu  DholoHu) of unfolding (Layton and Hellinga, 2010). At physiological temperature (310 K) the observed difference (DDG = 1.23 kcal mol1) translates into a roughly sevenfold higher binding affinity for oxido-squalene (KD = 5.44 lM) compared with squalene (KD = 40.10 lM) to SPFs. However, it should be emphasized that microDSC thermodynamic shifts were assessed in aqueous buffer. Under these conditions the free ligands are highly insoluble and are unlikely to dissociate from the denatured protein. Residual affinity of either ligand to denatured SPFs would thus impair the stability of the corresponding native SPFs ligand complex under equilibrium conditions and thus effect binding affinities of the SPFs ligand complexes in vitro.

3.5. Biological implications Cholesterol accounts for roughly half the molar lipid content in the plasma membrane of a mammalian cell (Bretscher and Munro, 1993). However, cholesterol is synthesized in the endoplasmic reticulum (ER), which has to exhibit five times lower cholesterol levels than the plasma membrane to remain deformable and permissive to membrane protein insertion (Mouritsen and Zuckermann, 2004). Therefore, mammalian cells are forced not only to continuously remove cholesterol from the ER but also to control the maintenance of cholesterol homeostasis through tight regulation of the key rate limiting enzymes HMG CoA reductase and squalene monooxygenase (Iyer et al., 1999). Similarly, the levels of the noncyclic cholesterol precursors squalene and

Fig.7. Schematic representation of the committed cholesterol synthesis pathway. The enzymes dedicated to the production of specific sterol end products are aligned along the ER membrane. Putative roles of SPF at the branching point between secretion of squalene and sterol endosynthesis as well as for the final step in the formation of the sterol backbone are highlighted in green.

2,3-oxidosqualene are tightly controlled to prevent their accumulation in the ER bilayer (Fig. 7). SPF facilitates the access of squalene to the monooxygenase active site (Friedlander et al., 1980), at the same time up-regulating 2,3-oxidosqualene (Shibata et al., 2001). To our knowledge, there has been no proof for direct interactions of SPF with membrane enzymes of the endogenous cholesterol synthesis to date. While it is certainly clear that SPF is able to stimulate squalene monooxygenase, the binding promiscuity of SPF for other lipids such as phosphatidylinositol may temper the conclusion that the natural, or only ligands of SPF are squalene and 2,3-oxidosqualene (Porter, 2003). We propose here, that SPF may facilitate substrate flow to and from the squalene monooxygenase. The slightly preferential binding of SPFs to oxidosqualene may indicate that other factors drive the reactions within the pathway forward. For instance, the oxidosqualene-lanosterol cyclase provides a sink to the endosynthetic pathway through its highly

Please cite this article in press as: Christen, M., et al. Structural insights on cholesterol endosynthesis: Binding of squalene and 2,3-oxidosqualene to supernatant protein factor. J. Struct. Biol. (2015), http://dx.doi.org/10.1016/j.jsb.2015.05.001

M. Christen et al. / Journal of Structural Biology xxx (2015) xxx–xxx

exergonic multi-step cyclization reaction converting the epoxide into lanosterol (Poralla et al., 1994). Another important function of SPF may emerge from its reported presence in human plasma (PAXDB integrated dataset (Wang et al., 2012), ID: 8653–8672). In line with this observation, two independent research groups reported highest SPF levels in northern blots of tissues with secretory activity such as liver, intestine, brain, skin and prostate (Shibata et al., 2001; Zimmer et al., 2000). It is thus possible that SPF may be responsible for the export of freshly synthesized squalene out of the committed sterol pathway towards other places where it has an active biological role (Fig. 7). Along these considerations, it has been recently shown by in vitro by pull down experiments, that SPF and the GOLGI resident protein transport regulator ACBD3 are both physically interacting partners of metal-dependent protein phosphatase (PPM1L) at the ER–Golgi membrane contact sites (ERGIC) (Shinoda et al., 2012). These findings clearly point towards a role for SPF in secretion since ERGIC is the site of segregation of secretory proteins, via packaging into COPII coated secretory transport vesicles (Schekman and Orci, 1996). In the rodent SPF knockout model chow-fed Spf/ mice were shown to suffer from decreased VLDL secretion and circulating cholesterol levels during fasting compared to in wild-type mice (Shibata et al., 2006). Interestingly, expression of the Sec14l3 gene that encodes the 45-kDa secretory protein (SPF2) with high similarity to SPF (86%), was reported to be inversely associated in vivo with the progression of experimentally induced airway inflammation of alveolar epithelium (Shan et al., 2009). It was further shown, that normal levels of the Sec14l3 mRNA positively correlated with the differentiation of ciliated epithelial cells and with airway integrity (Shan et al., 2012). Considering the proposed role of SPF (SEC14L2) in the committed cholesterol pathway and in secretion, our data may contribute to the molecular understanding of this crucial processes and the high-resolution structures may allow for the rational design of small molecule compounds with potential therapeutical use. 4. Conclusions Here we describe, for the first time, high-resolution X-ray structures of the SEC14-like domain of SPF in complex with squalene and 2,3-oxidosqualene. The removal of a self-inhibitory jelly-roll motif of SPF has proven to favor ligand binding and allowed us to experimentally assess the biophysical properties of the complexes. Structural analysis revealed structural constraints and micro-solvation features to be responsible for ligand selectivity. In fact, both computational modeling and thermal denaturation assays confirmed that both ligands are good SPF substrates with similar binding affinities. 5. PDB accession numbers The atomic coordinates and structure factors have been deposited in the Protein Data Bank, www.rcsb.org (PDB codes 4OMK, and 4OMJ). Acknowledgments This work was supported by the Swiss National Science Foundation through Grant N. 31003A_130497 and 31003A_ 156419 (A.S.) and Marie Heim-Voegtlin Grant N. PMPDP3_139609 (M.J.M.), and by the Norwegian Research Council through the CoE Centre for Theoretical and Computational Chemistry (CTCC) Grant Nos. 179568/V30 and 171185/V30 (M.C.a). We are grateful to Sandro Waltersperger and Vincent Olieric at the Swiss Light Source for their support in data collection. We thank Natacha

9

Olieric for her assistance and the groups of PD Dr. S. Schürch and Prof P. Renaud, for performing ESI-MS and GC–MS, respectively. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jsb.2015.05.001. References Adams, P.D., Afonine, P.V., Bunkoczi, G., Chen, V.B., Davis, I.W., Echols, N., Headd, J.J., Hung, L.W., Kapral, G.J., Grosse-Kunstleve, R.W., McCoy, A.J., Moriarty, N.W., Oeffner, R., Read, R.J., Richardson, D.C., Richardson, J.S., Terwilliger, T.C., Zwart, P.H., 2010. PHENIX: a comprehensive python-based system for macromolecular structure solution. Acta Crystallogr. D 66, 213–221. Anantharaman, V., Aravind, L., 2002. The GOLD domain, a novel protein module involved in Golgi function and secretion. Genome Biol. 3, research 0023. Aqvist, J., 1990. Ion water interaction potentials derived from free-energy perturbation simulations. J. Phys. Chem. U.S. 94, 8021–8024. Bloch, K.E., 1983. Sterol structure and membrane function. CRC Crit. Rev. Biochem. 14, 47–92. Bolze, C.S., Helbling, R.E., Owen, R.L., Pearson, A.R., Pompidor, G., Dworkowski, F., Fuchs, M.R., Furrer, J., Golczak, M., Palczewski, K., Cascella, M., Stocker, A., 2014. Human cellular retinaldehyde-binding protein has secondary thermal 9-cisretinal isomerase activity. J. Am. Chem. Soc. 136, 137–146. Bretscher, M.S., Munro, S., 1993. Cholesterol and the Golgi apparatus. Science 261, 1280–1281. Bruylants, G., Wouters, J., Michaux, C., 2005. Differential scanning calorimetry in life science. Thermodynamics, stability, molecular recognition and application in drug design. Curr. Med. Chem. 12, 2011–2020. Cascella, M., Barfuss, S., Stocker, A., 2013. Cis-retinoids and the chemistry of vision. Arch. Biochem. Biophys. 539, 187–195. Chin, J., Bloch, K., 1985. Stimulation by unsaturated fatty acid of squalene uptake in rat liver microsomes. J. Lipid Res. 26, 819–823. Chugh, A., Ray, A., Gupta, J.B., 2003. Squalene epoxidase as hypocholesterolemic drug target revisited. Prog. Lipid Res. 42, 37–50. Cieplak, P., Kollman, P.A., 1993. Peptide mimetics as enzyme inhibitors: use of free energy perturbation calculations to evaluate isosteric replacement for amide bonds in a potent HIV protease inhibitor. J. Comput. Aided Mol. Des. 7, 291–304. Cooper, A., 1999. Thermodynamic analysis of biomolecular interactions. Curr. Opin. Chem. Biol. 3, 557–563. D’angelo, I., Welti, S., Bonneau, F., Scheffzek, K., 2006. A novel bipartite phospholipid-binding module in the neurofibromatosis type 1 protein. EMBO Rep. 7, 174–179. DeLano, W.L., 2002. The PyMOL Molecular Graphics System. DeLano Scientific, San Carlos, CA, USA. Dietschy, J.M., Wilson, J.D., 1970. Regulation of cholesterol metabolism. I. N. Eng. J. Med. 282, 1128–1138. Emsley, P., Lohkamp, B., Scott, W.G., Cowtan, K., 2010. Features and development of Coot. Acta Crystallogr. D 66, 486–501. Essmann, U., Perera, L., Berkowitz, M.L., Darden, T., Lee, H., Pedersen, L.G., 1995. A smooth particle mesh Ewald method. J. Chem. Phys. 103, 8577–8593. Evans, P., 2006. Scaling and assessment of data quality. Acta Crystallogr. D 62, 72– 82. Feller, S.E., Zhang, Y.H., Pastor, R.W., Brooks, B.R., 1995. Constant pressure molecular dynamics simulation – the Langevin piston method. J. Chem. Phys. 103, 4613– 4621. Ferguson, J.B., Bloch, K., 1977. Purification and properties of a soluble protein activator of rat liver squalene epoxidase. J. Biol. Chem. 252, 5381–5385. Friedlander, E.J., Caras, I.W., Lin, L.F., Bloch, K., 1980. Supernatant protein factor facilitates intermembrane transfer of squalene. J. Biol. Chem. 255, 8042–8045. Greenfield, N.J., 2006. Using circular dichroism collected as a function of temperature to determine the thermodynamics of protein unfolding and binding interactions. Nat. Protoc. 1, 2527–2535. Hauss, T., Dante, S., Dencher, N.A., Haines, T.H., 2002. Squalane is in the midplane of the lipid bilayer: implications for its function as a proton permeability barrier. Biochim. Biophys. Acta 1556, 149–154. He, X.Q., Lobsiger, J., Stocker, A., 2009. Bothnia dystrophy is caused by domino-like rearrangements in cellular retinaldehyde-binding protein mutant R234W. Proc. Natl. Acad. Sci. U.S.A. 106, 18545–18550. Helbling, R.E., Aeschimann, W., Simona, F., Stocker, A., Cascella, M., 2012. Engineering tocopherol selectivity in alpha-TTP: a combined in vitro/in silico study. PLoS ONE 7, e49195. Helbling, R.E., Bolze, C.S., Golczak, M., Palczewski, K., Stocker, A., Cascella, M., 2013. Cellular retinaldehyde binding protein-different binding modes and microsolvation patterns for high-affinity 9-cis- and 11-cis-retinal substrates. J. Phys. Chem. B 117, 10719–10729. Helbling, R.E., Lamprakis, C., Aeschimann, W., Bolze, C., Stocker, A., Cascella, M., 2014. Mechanisms of ligand-protein interaction in Sec-14 like transporters investigated by computer simulations. Chimia 68, 615–619. Hess, B., Kutzner, C., van der Spoel, D., Lindahl, E., 2008. GROMACS 4: algorithms for highly efficient, load-balanced, and scalable molecular simulation. J. Chem. Theory Comput. 4, 435–447.

Please cite this article in press as: Christen, M., et al. Structural insights on cholesterol endosynthesis: Binding of squalene and 2,3-oxidosqualene to supernatant protein factor. J. Struct. Biol. (2015), http://dx.doi.org/10.1016/j.jsb.2015.05.001

10

M. Christen et al. / Journal of Structural Biology xxx (2015) xxx–xxx

Hoover, W.G., 1985. Canonical dynamics – equilibrium phase–space distributions. Phys. Rev. A 31, 1695–1697. Hornak, V., Abel, R., Okur, A., Strockbine, B., Roitberg, A., Simmerling, C., 2006. Comparison of multiple Amber force fields and development of improved protein backbone parameters. Proteins 65, 712–725. . Ikonen, E., 2008. Cellular cholesterol trafficking and compartmentalization. Nat. Rev. Mol. Cell Biol. 9, 125–138. Iyer, V.R., Eisen, M.B., Ross, D.T., Schuler, G., Moore, T., Lee, J.C., Trent, J.M., Staudt, L.M., Hudson Jr., J., Boguski, M.S., Lashkari, D., Shalon, D., Botstein, D., Brown, P.O., 1999. The transcriptional program in the response of human fibroblasts to serum. Science 283, 83–87. Jorgensen, W.L., Chandrasekhar, J., Madura, J.D., Impey, R.W., Klein, M.L., 1983. Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 79, 926–935. Kabsch, W., 2010. XDS. Acta Crystallogr. D 66, 125–132. Kollman, P., 1993. Free-energy calculations – applications to chemical and biochemical phenomena. Chem. Rev. 93, 2395–2417. Layton, C.J., Hellinga, H.W., 2010. Thermodynamic analysis of ligand-induced changes in protein thermal unfolding applied to high-throughput determination of ligand affinities with extrinsic fluorescent dyes. Biochemistry 49, 10831–10841. Leslie, A.G.W., Powell, H.R., 2007. Processing diffraction data with MOSFLM. Natl. Sci. Ser. II Math. 245, 41–51. Martyna, G.J., Klein, M.L., Tuckerman, M., 1992. Nose–Hoover chains – the canonical ensemble via continuous dynamics. J. Chem. Phys. 97, 2635–2643. Maxfield, F.R., Tabas, I., 2005. Role of cholesterol and lipid organization in disease. Nature 438, 612–621. Meier, R., Tomizaki, T., Schulze-Briese, C., Baumann, U., Stocker, A., 2003. The molecular basis of vitamin E retention: Structure of human alpha-tocopherol transfer protein. J. Mol. Biol. 331, 725–734. Min, K.C., Kovall, R.A., Hendrickson, W.A., 2003. Crystal structure of human alphatocopherol transfer protein bound to its ligand: implications for ataxia with vitamin E deficiency. Proc. Natl. Acad. Sci. U.S.A. 100, 14713–14718. Mokashi, V., Porter, T.D., 2005. Supernatant protein factor requires phosphorylation and interaction with Golgi to stimulate cholesterol synthesis in hepatoma cells. Arch. Biochem. Biophys. 435, 175–181. Mouritsen, O.G., Zuckermann, M.J., 2004. What’s so special about cholesterol? Lipids 39, 1101–1113. Nose, S., 1984. A Molecular-dynamics method for simulations in the canonical ensemble. Mol. Phys. 52, 255–268. Pettersen, E.F., Goddard, T.D., Huang, C.C., Couch, G.S., Greenblatt, D.M., Meng, E.C., Ferrin, T.E., 2004. UCSF chimera – a visualization system for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612. Poralla, K., Hewelt, A., Prestwich, G.D., Abe, I., Reipen, I., Sprenger, G., 1994. A specific amino acid repeat in squalene and oxidosqualene cyclases. Trends Biochem. Sci. 19, 157–158. Porter, T.D., 2003. Supernatant protein factor and tocopherol-associated protein: an unexpected link between cholesterol synthesis and vitamin E (review). J. Nutr. Biochem. 14, 3–6. Rostkowski, M., Olsson, M.H., Sondergaard, C.R., Jensen, J.H., 2011. Graphical analysis of pH-dependent properties of proteins predicted using PROPKA. BMC Struct. Biol. 11, 6. Ryan, M.M., Temple, B.R.S., Phillips, S.E., Bankaitis, V.A., 2007. Conformational dynamics of the major yeast phosphatidylinositol transfer protein Sec14p: insight into the mechanisms of phospholipid exchange and diseases of Secl4plike protein deficiencies. Mol. Biol. Cell 18, 1928–1942. Ryckaert, J.P., Ciccotti, G., Berendsen, H.J.C., 1977. Numerical-integration of Cartesian equations of motion of a system with constraints – moleculardynamics of n-alkanes. J. Comput. Phys. 23, 327–341. Schaaf, G., Ortlund, E.A., Tyeryar, K.R., Mousley, C.J., Ile, K.E., Garrett, T.A., Ren, J., Woolls, M.J., Raetz, C.R.H., Redinbo, M.R., Bankaitis, V.A., 2008. Functional

anatomy of phospholipid binding and regulation of phosphoinositide homeostasis by proteins of the Sec14 superfamily. Mol. Cell 29, 191–206. Schekman, R., Orci, L., 1996. Coat proteins and vesicle budding. Science 271, 1526– 1533. Sha, B., Phillips, S.E., Bankaitis, V.A., Luo, M., 1998. Crystal structure of the Saccharomyces cerevisiae phosphatidylinositol-transfer protein. Nature 391, 506–510. Shan, L., Noritake, S., Fujiwara, M., Asano, S., Yoshida-Noro, C., Noro, N., Yamashita, K., Kawakami, T., 2012. Sec14l3 is specifically expressed in mouse airway ciliated cells. Inflammation 35, 702–712. Shan, L., Kawakami, T., Asano, S., Noritake, S., Yoshimoto, D., Yamashita, K., Kikkawa, H., Kinoshita, M., Matsubara, S., 2009. Inverse relationship between Sec14l3 mRNA/protein expression and allergic airway inflammation. Eur. J. Pharmacol. 616, 293–300. Shibata, N., Arita, M., Misaki, Y., Dohmae, N., Takio, K., Ono, T., Inoue, K., Arai, H., 2001. Supernatant protein factor, which stimulates the conversion of squalene to lanosterol, is a cytosolic squalene transfer protein and enhances cholesterol biosynthesis. Proc. Natl. Acad. Sci. U.S.A. 98, 2244–2249. Shibata, N., Jishage, K., Arita, M., Watanabe, M., Kawase, Y., Nishikawa, K., Natori, Y., Inoue, H., Shimano, H., Yamada, N., Tsujimoto, M., Arai, H., 2006. Regulation of hepatic cholesterol synthesis by a novel protein (SPF) that accelerates cholesterol biosynthesis. FASEB J. 20, 2642–2644, official publication of the Federation of American Societies for Experimental Biology. Shinoda, Y., Fujita, K., Saito, S., Matsui, H., Kanto, Y., Nagaura, Y., Fukunaga, K., Tamura, S., Kobayashi, T., 2012. Acyl-CoA binding domain containing 3 (ACBD3) recruits the protein phosphatase PPM1L to ER–Golgi membrane contact sites. FEBS Lett. 586, 3024–3029. Stocker, A., Baumann, U., 2003. Supernatant protein factor in complex with RRRalpha-tocopherylquinone: a link between oxidized Vitamin E and cholesterol biosynthesis. J. Mol. Biol. 332, 759–765. Stocker, A., Zimmer, S., Spycher, S.E., Azzi, A., 1999. Identification of a novel cytosolic tocopherol-binding protein: structure, specificity, and tissue distribution. IUBMB Life 48, 49–55. Stocker, A., Tomizaki, T., Schulze-Briese, C., Baumann, U., 2002. Crystal structure of the human supernatant protein factor. Structure 10, 1533–1540. Studier, F.W., 2005. Protein production by auto-induction in high density shaking cultures. Protein Expr. Purif. 41, 207–234. Tai, H.H., Bloch, K., 1972. Squalene epoxidase of rat liver. J. Biol. Chem. 247, 3767– 3773. Tchen, T.T., Bloch, K., 1957. On the conversion of squalene to lanosterol in vitro. J. Biol. Chem. 226, 921–930. Wallace, A.C., Laskowski, R.A., Thornton, J.M., 1995. Ligplot – a program to generate schematic diagrams of protein ligand interactions. Protein Eng. 8, 127–134. Wang, J., Wolf, R.M., Caldwell, J.W., Kollman, P.A., Case, D.A., 2004. Development and testing of a general amber force field. J. Comput. Chem. 25, 1157–1174. Wang, M., Weiss, M., Simonovic, M., Haertinger, G., Schrimpf, S.P., Hengartner, M.O., von Mering, C., 2012. PaxDb, a database of protein abundance averages across all three domains of life. Mol. Cell. Proteomics MCP 11, 492–500. Welti, S., Fraterman, S., D’Angelo, I., Wilm, M., Scheffzek, K., 2007. The Sec14 homology module of neurofibromin binds cellular glycerophospholipids: mass spectrometry and structure of a lipid complex. J. Mol. Biol. 366, 551–562. Winn, M.D., Ballard, C.C., Cowtan, K.D., Dodson, E.J., Emsley, P., Evans, P.R., Keegan, R.M., Krissinel, E.B., Leslie, A.G.W., McCoy, A., McNicholas, S.J., Murshudov, G.N., Pannu, N.S., Potterton, E.A., Powell, H.R., Read, R.J., Vagin, A., Wilson, K.S., 2011. Overview of the CCP4 suite and current developments. Acta Crystallogr. D 67, 235–242. Yamamoto, S., Bloch, K., 1970. Studies on squalene epoxidase of rat liver. J. Biol. Chem. 245, 1670–1674. Zimmer, S., Stocker, A., Sarbolouki, M.N., Spycher, S.E., Sassoon, J., Azzi, A., 2000. A novel human tocopherol-associated protein: cloning, in vitro expression, and characterization. J. Biol. Chem. 275, 25672–25680.

Please cite this article in press as: Christen, M., et al. Structural insights on cholesterol endosynthesis: Binding of squalene and 2,3-oxidosqualene to supernatant protein factor. J. Struct. Biol. (2015), http://dx.doi.org/10.1016/j.jsb.2015.05.001