Biochemical Journal

Research article

Cholesterol binding is a prerequisite for the activity of the steroidogenic acute regulatory protein (StAR)

Alireza Roostaee, Élie Barbar, Jean-Guy LeHoux, Pierre Lavigne


Steroidogenesis depends on the delivery of cholesterol from the outer to the inner mitochondrial membrane by StAR (steroidogenic acute regulatory protein). However, the mechanism by which StAR binds to cholesterol and its importance in cholesterol transport are under debate. According to our proposed molecular model, StAR possesses a hydrophobic cavity, which can accommodate one cholesterol molecule. In the bound form, cholesterol interacts with hydrophobic side-chains located in the C-terminal α-helix 4, thereby favouring the folding of this helix. To verify this model experimentally, we have characterized the in vitro activity, overall structure, thermodynamic stability and cholesterol-binding affinity of StAR lacking the N-terminal 62 amino acid residues (termed N-62 StAR). This mature form is biologically active and has a well-defined tertiary structure. Addition of cholesterol to N-62 StAR led to an increase in the α-helical content and T° (melting temperature), indicating the formation of a stable complex. However, the mutation F267Q, which is located in the C-terminal helix interface lining the cholesterol-binding site, reduced the biological activity of StAR. Furthermore, the cholesterol-induced thermodynamic stability and the binding capacity of StAR were significantly diminished in the F267Q mutant. Titration of StAR with cholesterol yielded a 1:1 complex with an apparent KD of 3×10−8. These results support our model and indicate that StAR can readily bind to cholesterol with an apparent affinity that commensurates with monomeric cholesterol solubility in water. The proper function of the C-terminal α-helix is essential for the binding process.

  • cholesterol
  • circular dichroism
  • mitochondria
  • steroidogenesis
  • steroidogenic acute regulatory protein (StAR)
  • thermodynamics


StAR (steroidogenic acute regulatory protein) is the critical mediator of de novo synthesis of adrenal and gonadal steroids by the transfer of cholesterol from the OMM (outer mitochondrial membrane) to the IMM (inner mitochondrial membrane) where it is processed to pregnenolone by the action of cytochrome P450scc (scc is side chain cleavage) [1,2]. Transport of cholesterol across the mitochondrial membrane appears to be the rate-limiting step in steroidogenesis [3,4] and is critically influenced by trophic hormone-induced StAR expression [5,6]. In humans, StAR is synthesized as a 285 amino acid proprotein in the cytoplasm with an apparent molecular mass of 37 kDa. Upon translocation into mitochondria, it is processed to yield a mature 30 kDa protein [7]. The acute expression of StAR in the adrenal glands and gonads is modulated by a trophic hormone-induced generation of cAMP with subsequent phosphorylation and gene transcription [8,9]. The physiological importance of StAR was confirmed in studies revealing that LCAH (lipoid congenital adrenal hyperplasia) resulted from StAR abnormalities [10]. In LCAH, cholesterol cannot be metabolized to pregnenolone and cortisol synthesis is interrupted, causing an increase of ACTH (corticotropin) release. This trophic hormone stimulates cholesterol recruitment and its accumulation as an ester derivative in lipid droplets of steroidogenic cells [10,11]. Thirty mutations in the StAR gene are known to cause LCAH. Most of these mutations are located between exons 5 and 7 and affect the StART (StAR-related lipid transfer) domain, in particular the critical C-terminal region [12].

StAR is mainly composed of two functional domains: a mitochondrial targeting sequence (N-terminal 62 amino acids) which is required for its translocation into the mitochondria and a StART domain which is believed to bind and transfer cholesterol across the mitochondrial membrane [13]; StAR is rapidly degraded after entering the mitochondria [14]. It is not clear how the protein facilitates cholesterol transfer from the OMM to the IMM. There is evidence that the StAR protein lacking the N-terminal targeting sequence (termed N-62 StAR) can stimulate steroidogenesis in transfected COS-1 cells [15] and steroidogenic mitochondria in vitro [16]. Apart from the importance of the protein targeting to mitochondria for its steroidogenic activity [17], there is a general consensus that StAR acts on the OMM to promote cholesterol transfer to the IMM [18]. However, it is not clearly established whether StAR binds to cholesterol in the cytosol or extracts it from the OMM [19]. Moreover, the molecular mechanism of cholesterol binding to StAR is still under debate. It has been suggested that the binding and release of cholesterol could depend on the existence of a molten globular state of StAR that is generated proximal to the mitochondrial membrane where the pH is presumably below 4.0 [20], although such an acidic environment in the cell is yet to be demonstrated. By structural homology modelling, based on the crystallographic structure of the MLN-64-StART domain [21], we have suggested that the cholesterol-binding site of StAR is a well-ordered hydrophobic cavity containing a salt bridge which is important for the specificity of the interaction with cholesterol (PDB entry: 1IMG) [22]. In addition, we propose that the entry (and dissociation) of cholesterol into (and from) this cavity is gated by a reversible unfolding/refolding of the C-terminal α-helix [22] and that variations in the size of the cavity provide the driving force for the reversible unfolding of the C-terminal α-helix. This model suggests that in the absence of cholesterol, the fully folded state of StAR is in equilibrium with a partially folded state that allows for the entry and exit of cholesterol, and that the binding of cholesterol to the cavity stabilizes the fully folded state as well as the C-terminal α-helix.

To shed light on the processing of cholesterol transfer by StAR to the IMM, we sought to clarify the binding mechanism of StAR to cholesterol, hypothesizing that this might be the first step in the transfer mechanism. To provide experimental validation of our model, we expressed, purified and measured the steroidogenic activity of N-62 StAR. In addition, the secondary structure and the thermodynamic stability of N-62 StAR in the presence or absence of cholesterol were characterized by CD. Finally, the stoichiometry and the affinity (KD) of N-62 StAR for cholesterol were determined by monitoring the cholesterol-induced secondary structure changes of the protein. A StAR mutant with a mutated C-terminal α-helix (F267Q) was also studied. The results of the present study support our model for cholesterol binding. This constitutes an important step toward understanding the elusive mechanism of cholesterol transport from the cytoplasm to the IMM.


Restriction enzymes were purchased from New England Biolabs. Human StAR cDNA and oligonucleotides were purchased from Invitrogen. Cloning vectors (pET-21b and pET-3a) were purchased from Novagen. XLI Blue Escherichia coli competent cells were obtained from Stratagene. [1,1,6,7]-Pregnenolone (95.0 Ci/mmol) was purchased from Amersham whereas 22(R)-hydroxycholesterol and cholesterol were purchased from Sigma–Aldrich. Nickel-affinity resin was purchased from Qiagen. Pregnenolone rabbit antiserum was obtained from ICN Biochemicals. All other chemicals were purchased from local suppliers.

Cloning, expression and purification of His6-tagged N-62 StAR

Recombinant N-62 StAR was amplified by PCR from the human StAR full-length cDNA while a His6 tag was being added at the C- or N-terminus, and then cloned into PET-3a and PET-21b vectors respectively. Hereafter, these StAR constructs will be referred to as C-StAR and N-StAR. An F267Q mutant was generated by PCR using N-StAR as a template. This construct will be referred to as N-StAR F267Q. PCR amplification fidelity was confirmed by sequencing. PCR primers and corresponding cloning sites are listed in Supplementary Table S1 at Constructs were overexpressed in BL-21 (DE3) cells at 25 °C using 0.1 mM IPTG (isopropyl β-D-thiogalactoside) for 5 h. The bacterial pellets were lysed in 300 mM NaCl, 50 mM NaH2PO4 (pH 8.0), 1 mM PMSF, 1 mM TCEP [tris-(2-carboxyethyl)phosphine] and 1 mg/ml lysozyme, followed by drawing–expelling of the lysates through a 21-gauge-needle and centrifugation for 45 min at 20000 g. Purification of the protein from the supernatant was carried out using nickel-affinity chromatography according to the manufacturer's protocol (Qiagen). Proteins were eluted in a buffer consisting of 125 mM NaCl, 300 mM imidazole, 1 mM TCEP, 25 mM NaH2PO4 (pH 7.4). This buffer was then exchanged for a potassium phosphate buffer consisting of 50 mM K3PO4, 2 mM KCl, 70 mM acetic acid, 9 mM KH3PO4 and 5 mM Na2HPO4 (pH 7.4), using Millipore ultracentrifugation filters (Amicon Canada). The protein concentration was estimated by the absorbance at 280 nm, considering a molar absorption coefficient of 27080 M−1·cm−1 for N-62 StAR including a His6 tag.

Assessment of StAR steroidogenic activity in the presence of purified mitochondria

Mitochondria were isolated from mouse MA-10 and human Y-1 steroidogenic cells as previously described [23] and were used immediately or stored at −80 °C [24]. To determine the steroidogenic activity of StAR in vitro [16], 10 μM StAR was incubated with 1 μg/μl mitochondria and 200 μM cholesterol in a modified buffer consisting of 5 mM sodium succinate and 1 μM epostane (3β-hydroxysteroid dehydrogenase inhibitor; pH 7.4). Incubations were carried out at 37 °C for 90 min. Reactions were terminated by flash-freezing and the amount of pregnenolone produced was quantified by radioimmunoassay [25,26]. Results are expressed as the mean pregnenolone concentration (ng of pregnenolone/μg of mitochondrial protein).

StAR radioligand-binding assay

Binding assays were performed using [3H]cholesterol as the radiolabelled ligand. Purified protein (2.5 μg/ml) was incubated for 60 min with 1 pmol of [3H]cholesterol in the absence or presence of 5 μM unlabelled cholesterol in 100 μl of PBS. Some binding experiments were performed at different pH levels ranging from 2.5 to 7.4. After 60 min incubation, the volume was increased to 500 μl by the addition of 100 mM PBS containing 0.3% gelatin. To distinguish the effect of dilution on the StAR–cholesterol dissociation, some binding experiments were performed in a total volume of 500 μl of PBS. Prior to centrifugation (830 g for 15 min at 4 °C), 0.01% Triton X-100 was added to reduce non-specific binding of proteins to the tubes. Bound and free cholesterol were separated as previously described [25]. The radioactivity of bound [3H]cholesterol was measured with a Beckman LS8000 CE Scintillation Counter (Beckman Coulter). Background levels were subtracted from the total counts and data were analysed using SigmaPlot software (Systat Software).

CD spectroscopy

CD measurements were carried out as previously described [27] using a protein concentration of 0.16 mg/ml in a 25 mM phosphate buffer at pH 7.4. Cholesterol-induced changes in the secondary structure were evaluated by monitoring the ellipticity in the far-UV region (190–250 nm). Protein thermal denaturation curves were determined in the absence or presence of cholesterol at pH 7.4 by monitoring the CD values at 222 nm, with increasing temperature from 10 °C to 90 °C. In some experiments, StAR was pre-incubated with cholesterol for 45 min. The temperature-induced denaturation curves monitored by CD were acquired as described elsewhere [27]. The temperature-dependence of the mean residue ellipticity (θ) at 222 nm was fitted using an in-house non-linear least-squares fitting program [28].

Data analysis

In order to obtain the thermodynamic parameters [T°, ΔHu°(T°), ΔSu°(T°), ΔGu°(T°) and Pu(T)] describing the stability of StAR in the absence or presence of cholesterol, the thermal denaturation curves of StAR were simulated assuming a two-state unfolding mechanism [29,30]. The observed value of θ at any temperature is given by θ=θFPFUPU, where θF and θU are the values of θ in the folded and unfolded states respectively. Following this approach [27], the temperature-dependence of the molar ellipticity ([θ]M) is given by eqn (1): Embedded Image(1) where ΘN(T) and ΘU(T) are the temperature-dependent molar ellipticities (baselines) for the native and unfolded states respectively, assuming the following linear equations (eqns 2 and 3): Embedded Image(2) Embedded Image(3) where ΘN(0) and ΘU(0) are the mean residue ellipticities at 0 °C for the folded and unfolded state respectively and dΘN(T)/dT and dΘU(T)/dT are the constant slopes of ΘN(T).

The population of the unfolded state, is given by eqn (4): Embedded Image(4) where R is the gas constant and T is the absolute temperature. Therefore in order to obtain Pu(T), ΔGu(T) must be calculated at all temperatures. As described previously [28], this can be achieved by the determination of T°, the melting temperature, and ΔHu°(T°) the enthalpy of unfolding at T° through least-squares fitting, assuming the following expression for ΔGu(T): Embedded Image(5) In the present study, the heat capacity of unfolding ΔCp,u was considered constant and equal to 1 kcal·mol−1·K−1 (1 kcal≈4.184 kJ).

Statistical analysis was performed by ANOVA and P values less than 0.05 were considered significant.

CD-monitored cholesterol titration

For the stoichiometric and binding analysis, titration was carried out using increasing amounts of cholesterol solubilized in ethanol, from 0 to 200% relative to the total protein concentration (6 μM) in a 25 mM phosphate buffer (pH 7.4). After 90 min equilibrium time, samples were analysed by CD at 25 °C. The apparent KD (KD*) was obtained by fitting the following equation (eqn 6) [31]: Embedded Image(6) where P(M) is the total molar protein concentration, C(M) is the total molar cholesterol concentration and KD (1/KA) is the dissociation constant [32]. All titration experiments were repeated three to five times and the mean CD signal changes were plotted against the concentration of total cholesterol in solution. The α-helical contents of N-62 StAR were also calculated from titration experiments using the method described by Adler et al. [33], which was based on far-UV data at 208 and 222 nm at 37 °C.


Assessment of N-62 StAR functionality: steroidogenic activity

Previous studies have shown that the addition of a histidine tag to the N- or C-terminus of the N-62 StAR protein did not affect its steroidogenic activity [16] while retaining its biological activity [15]. Thus we used N- and C-StAR in our experiments.

The steroidogenic activity of StAR was evaluated by determination of pregnenolone production using mitochondria isolated from Y-1 and Leydig MA-10 cells [16]. Figure 1 shows that mitochondria from Y-1 and MA-10 cells produced 0.21±0.03 and 0.12±0.01 ng of pregnenolone per μg of protein respectively: this is the basal pregnenolone production without the addition of cholesterol or StAR and may be due to the traces of endogenous cholesterol found in the OMM. Then, the functionality of the mitochondria was assessed by measuring the production of pregnenolone from 22(R)-hydroxycholesterol, a soluble cholesterol analogue that permeates the OMM. 22(R)-hydroxycholesterol maximally increased Y-1 and MA-10 mitochondrial pregnenolone production by approx. 40-fold when compared with the buffer control (Figure 1). Under the same conditions, mitochondria disrupted by freeze–thawing showed no increase in the pregnenolone level. These results confirm that the mitochondria used in the present study were active and functional.

Figure 1 Steroidogenic activity of N-StAR and N-StAR F267Q mutant

The steroidogenic activity of N-StAR and C-StAR was assessed for mitochondria isolated from Leydig MA-10 (grey columns) or Y-1 (black columns) cells. Mitochondria (1 μg/μl) were incubated for 90 min at 37 °C with buffer or 22(R)-hydroxycholesterol (22R-Ch; control experiments), or with different combination of StAR proteins (10 μM) with/without 200 μM cholesterol (Ch). Pregnenolone production was measured by radioimmunoassay. Values are expressed as the means±S.E.M. of pregnenolone produced per microgram of mitochondrial protein (ng/μg of protein) from three separate experiments. x, P<0.05, compared with the Y-1 buffer control; *, P<0.05, compared with MA-10 buffer control.

Addition of 200 μM cholesterol (exogenous cholesterol) did not induce any activity in either of the mitochondrial types. Also, the addition of N-StAR or C-StAR slightly enhanced pregnenolone production by MA-10 mitochondria (2.5±0.91- and 3.2±0.25-fold respectively). However, no significant increase was observed with Y-1 mitochondria. This may reflect differences between MA-10 and Y1 cell lines. In the two mitochondrial types, a higher pregnenolone production occurred following the addition of both cholesterol and C-StAR or N-StAR. C-StAR and N-StAR stimulated pregnenolone production in Y-1 mitochondria by 5.86±0.81- and 6.75±0.55-fold respectively, and in MA-10 mitochondria by 14.36±0.94- and 15.86±0.87-fold respectively. This stimulation of pregnenolone production by StAR in the presence of cholesterol was time-dependent and saturated after 90 min (Supplementary Figure S1 at Given that N-StAR and C-StAR gave similar activities, we proceeded with the remaining experiments of this work using N-StAR. The integrity of α-helix 4 is essential for the biological function of StAR [34]. In this regard, we investigated the steroidogenic activity of the α-helix-4-mutated N-StAR. Thus the N-StAR F267Q mutant increased Y-1 and MA-10 mitochondrial pregnenolone production by 2.11±0.48- and 4.96±0.89-fold respectively. However, the increase of pregnenolone production by the F267Q mutant was significantly lower than the increase induced by the wild-type N-StAR; this represents 28±2.9% and 26.23±3.4% of the induction by N-StAR in Y-1 and MA-10 respectively. Taken together, these results show that a high pregnenolone production by isolated mitochondria is achieved in the presence of both cholesterol and wild-type StAR.

Determination of the binding specificity of StAR to cholesterol

It has been suggested that a progressive decrease of the pH may exist in the close proximity of the OMM [20]. This was proposed to be important for StAR function. In this regard, we investigated the effect of pH on the formation of the StAR–cholesterol complex. This was performed using the radioligand-binding assay as described in the Experimental section. We observed that the binding of radiolabelled cholesterol to StAR was maximal at pH 7.4 (Supplementary Figure S2 at Therefore this pH value was selected for the following in vitro experiments. To determine the extent of the binding specificity of cholesterol to StAR, competitive/displacement binding assays were performed with unlabelled cholesterol as a competitor (Figure 2). The addition of unlabelled cholesterol to the StAR–[3H]cholesterol complex resulted in a 0.136±0.035 pmol displacement of bound [3H]cholesterol which corresponded to more than 50% of the total amount of bound [3H]cholesterol (P<0.005); this was indicative of the specificity of binding. These observations contrast with the low amount of cholesterol-binding capacity of the F267Q mutant and the non-significant displacement by unlabelled cholesterol (0.032±0.001 pmol, P=0.18) (Figure 2). Therefore the F267Q mutant does not form a stable complex with cholesterol, which correlates with its lower activity for the ligand, compared with the wild-type protein (Figure 1). These results suggest that a stable StAR–cholesterol interaction is necessary for both cholesterol binding and the steroidogenic activity of mitochondria.

Figure 2 Competitive binding of cholesterol to StAR at physiological pH

N-StAR or N-StAR F267Q (2.5 μg/ml) were incubated for 90 min with 1 pmol of [3H]cholesterol in PBS (pH 7.4) at 25 °C. Competition/displacement assays were performed by displacing the [3H]cholesterol with an excess of unlabelled cholesterol (1 μM). Ch, cholesterol; Ch*, [3H]cholesterol. Values are expressed as the means±S.E.M. from three separate experiments. *, P<0.05; NS, non significant.

Cholesterol induces folding of the C-terminal helix of StAR

According to the available molecular models of N-62 StAR [22,35], the secondary-structure content can be estimated to be approx. 35% α-helices and 43% β-strands in the fully folded state (cholesterol-loaded state). However, in the absence of cholesterol, we have proposed [22,35] that a hydrophobic cavity promotes the destabilization of the tertiary structure involving the C-terminal α-helix. Consequently, this α-helix undergoes a folding/unfolding transition in the apo-form. In fact, the reversible unfolding of this helix may facilitate the access (and dissociation) of cholesterol to (and from) its binding site. Binding of cholesterol to this state of StAR (unfolded helix) can eventually induce the refolding of helix, thereby stabilizing the StAR structure.

In order to obtain experimental data to validate this model, we have determined the secondary-structure content of N-62 StAR in the presence or absence of cholesterol. The far-UV CD spectra of N-62 StAR recorded at pH 7.4 in the absence of cholesterol is shown in Figure 3(A). The spectra depict a shallow double minimum at 208 and 222 nm and a maximum at ∼193 nm, as expected for proteins composed of α-helical and β-sheet secondary structures. Furthermore, in the presence of an equimolar quantity of cholesterol, the minima at 208 and 222 nm are more pronounced, indicative of the stabilization of the helical content (Figure 3A). The helical stabilization can be monitored by an isodichroic point near 203 nm [36], which was recorded upon cholesterol binding to StAR (Figure 3A, arrow). Of note, random coils and α-helices have the same molar ellipticities at 203 nm. Therefore it is expected that no ellipticity changes should occur at 203 nm if the C-terminal helix shifts from a partially random-coiled state to a stable helical state, in the presence of cholesterol.

Figure 3 Effect of cholesterol binding to StAR

CD spectra of (A) N-StAR or (B) N-StAR F267Q (6 μM) in the absence or presence of an equimolar concentration of cholesterol (6 μM) at pH 7.4. Spectra were taken at different time intervals from 0 to 90 min. Three spectra (representative of three different experiments) were selected and shown before addition of cholesterol (solid line), immediately after addition of cholesterol (broken line) and 90 min after the addition of cholesterol (dotted line). An isodichroic point (indicated by the arrow) occurs at 203 nm, indicating that cholesterol induces mainly coil-to-helix transitions at both 222 nm and 208 nm.

To infer the location to the C-terminal helix, we studied the F267Q mutant. According to the existing molecular models [22,35], Phe267 is located in the C-terminal α-helix and is involved in tertiary hydrophobic interactions in the fully folded apo form and contributes to the StAR–cholesterol complex formation. The replacement of Phe267 with a polar glutamine residue is expected to weaken both the apo-tertiary structure and the cholesterol–StAR interaction. Indeed, the F267Q mutant had a lesser amount of secondary-structure content (Figure 3B, solid line) as compared with wild-type protein (Figure 3A, solid line). Also, the addition of cholesterol to the F267Q mutant had only a slight effect on the secondary structure (Figure 3B, broken line) in comparison with the wild-type (Figure 3A, broken line). These observations are consistent with the interpretation of a significant contribution of the C-terminal α-helix to the overall structure of StAR. Additionally, the results confirm that binding of StAR to cholesterol can eventually lead to the refolding of α-helix 4. Finally, the helical content of StAR has been determined from the ellipticities at 208 and 222 nm in terms of percentage helicity, using the method described by Adler et al. [33]. In the absence of cholesterol, the CD spectrum of StAR possessed approx. 24.7% ±0.8% helicity. The helical content in the presence of cholesterol was increased by up to 33.9%±2.3% after equilibrium was reached (∼60 min). The increased volume of the helical structure of StAR is in close agreement with the value predicted from our model in its bound state [22]. However, consistent with the CD spectra, the approximate amount of helical content of F267Q was lower than that of the wild-type StAR (22.1%±0.5% and 24.5%±0.9% in the absence and presence of cholesterol respectively).

Binding of cholesterol improves the conformational stability of StAR

The presence of a ligand at the binding site of a protein typically stabilizes the conformation of the protein. Therefore we investigated whether this principle applied to StAR and cholesterol. In this context, changes in heat-induced conformational stability of StAR were monitored by CD at 222 nm ([θ]222) over a range 10–90 °C. The unfolding profile of StAR at physiological pH was associated with a single and co-operative transition (Figure 4A), indicating a well-defined tertiary structure and that StAR unfolding followed a two-step process of denaturation. A similar thermal denaturation of StAR has been previously reported [37]. However, the characterization of the detailed thermodynamics of unfolding and the influence of cholesterol on the thermodynamics of unfolding were not reported. In the present study, we show that the binding of cholesterol shifts the temperature denaturation curve of N-62 StAR to a higher value. A larger secondary structure (α-helix) content depicted by the more negative [θ]222 was also observed (Figure 4A, ●). A similar co-operative denaturation profile was observed in the case of the F267Q mutant (Figure 4B, ●). But, the F267Q mutant has a lower helical content and the addition of cholesterol induced only a slight increase of the helicity and T° of the mutant, compared with the wild-type protein.

Figure 4 Effect of cholesterol on the thermal stability of StAR monitored by CD

Unfitted molar ellipticity points ([θ]222) obtained from the thermal denaturation of (A) N-StAR and (B) N-StAR F267Q mutant. StAR (○) and StAR–cholesterol complex (●) recorded in 25 mM sodium phosphate at pH 7.4. Protein (6 μM) without or with an equimolar concentration of cholesterol was incubated at 25 °C for 90 min, then thermal denaturations were repeated at least three times.

Using an in-house non-linear least-squares fitting program, temperature-denaturation curves {[θ]222 (T)} were fitted to eqn (1), to obtain T°, ΔHu°(T°), ΔSu°(T°) and ΔGu°(T°). All of the thermodynamic parameters were measured from the equations described in the Experimental section. At pH 7.4, the T° increased from 42.3 °C in the absence of cholesterol to 45.9 °C at cholesterol saturation (Figure 5A, bottom panel). However, the T° of unfolding was lower for the F267Q mutant, with values of 40.5 °C and 41.6 °C in the absence or presence of cholesterol respectively (Figure 5B, bottom panel). As depicted in Figure 5(A) (upper panel) at 37 °C the free energy of unfolding of the bound state had a higher value [ΔGu° (37 °C)=2.43±0.07 kcal·mol−1] when compared with that of the apo state [ΔGu° (37 °C)=1.35±0.08 kcal·mol−1], with a ΔΔGu° (37 °C) =1.08±0.01 kcal·mol−1. The increase in free energy of unfolding clearly points out the favourable interaction between StAR and cholesterol which gives rise to a more stable conformation of the protein. For a better understanding of the nature of interaction between StAR and cholesterol, variations in enthalpy and entropy of unfolding were determined as a function of temperature. As reported in Table 1, the ΔHu° (37 °C) for StAR and StAR–cholesterol complex were 87.55±0.15 kcal·mol−1 and 79.80±0.15 kcal·mol−1 respectively. This leads to a ΔΔHu° (37 °C) of −7.75 kcal·mol−1 in the presence of cholesterol and −2.38 kcal·mol−1 in its absence. The results indicate that less favourable (enthalpic) interactions are broken during the unfolding of the complex. Knowing the corresponding ΔΔGu° (37 °C) (1.08 kcal·mol−1), the variation in entropy TΔΔSu° (37 °C) of unfolding of StAR caused by the presence of cholesterol was calculated to be −8.83 kcal·mol−1. This indicates that the unfolding of the complex is accompanied by a less favourable (less positive) entropy. This behaviour could be the result of the exposure of more buried hydrophobic surfaces upon unfolding. The corresponding thermodynamic parameters for the simulation of unfolding profiles of the F267Q mutant are also summarized in Table 1.

View this table:
Table 1 Thermodynamic parameters determined from the simulations of temperature-induced denaturation of StAR and the F267Q mutant in the presence or absence of cholesterol at 37 °C
Figure 5 Stability curve of StAR at physiological pH

Fitting of experimental data from Figure 4, using eqns (1), (4) and (5). The stability curves (upper panels) for (A) N-StAR and (B) N-StAR F267Q mutant without (dotted lines) and with (solid lines) cholesterol were calculated from a non-linear least-squares fitting using eqn (5), as described in the Experimental section. The lower panels represent the temperature-dependence of the population of unfolded state (Pu), for (A) N-StAR and (B) N-StAR F267Q mutant, which was calculated from eqn (4) and fitted to eqn (1). The parameters obtained for N-StAR are: T°=42.3 °C, ΔHu =90.4 kcal·mol−1, and those obtained for the N-StAR–cholesterol are: T°=45.9 °C and ΔHu=86.2 kcal·mol−1. Dotted and solid lines represent the variation of unfolded population (Pu) for N-StAR and the N-StAR–cholesterol complex respectively.

Ultimately, the stabilization of the bound state by cholesterol should be accompanied by a decrease in the population of the unfolded state (Pu) (Figure 5). For instance, at a physiological temperature (37 °C), the Pu of StAR in the absence of cholesterol is approx. 8% (Figure 5A, lower panel, solid line) and in the presence of an equimolar concentration of cholesterol, this population is decreased to approx. 2% (Figure 5A, lower panel, dotted line). Of significance, the Pu of the F267Q mutant was significantly increased (23%) in the absence of cholesterol. As shown in Figure 5(B) (lower panel), the binding of cholesterol resulted in a smaller decrease in the Pu of the mutant (18%) as compared with the wild-type protein.

StAR-cholesterol binding stoichiometry and dissociation constant (KD)

To unveil the mechanism underlying the StAR–cholesterol interaction, the apparent KD* and the stoichiometry of the complex formed between N-62 StAR and cholesterol were determined. Based on the induction of secondary structure upon addition of cholesterol to StAR, the change in [θ]222 was monitored as a function of the total cholesterol concentration. If the StAR–cholesterol interaction is specific and stable (with an apparent KD*<10−8), a stabilization of the α-helical structure should follow a typical binding pattern with a saturation at, or close to, the concentration leading to a stoichiometric complex. As observed, the titration of 6 μM protein with cholesterol resulted in a typical variation of Δ[θ]222 reaching a plateau value at a 1:1 binding (Figure 6). This indicates that StAR has only one specific binding site for cholesterol. A simulation of the saturation curve with eqn (6) gave a KD*∼3×10−8. Taking into consideration the poor solubility of cholesterol as a monodispersed species, the measured KD may be interpreted as an apparent (KD*). The cmc (critical micellar concentration) of cholesterol is ∼1×10−8 M [38] which implies that, in our binding assay, cholesterol exists in an equilibrium between a free monomeric state and a micellar structure (Figure 7B). Consequently, it is plausible that StAR may desorb cholesterol from the micelles or directly bind the free cholesterol. Considering that the apparent KD* has the same order of magnitude (10−8) as the cmc of cholesterol, both pathways may be possible in vitro.

Figure 6 Titration of StAR with cholesterol monitored by CD

The variation in [θ]222 (Δ[θ]222) reached after equilibrium is plotted as a function of the total concentration of the added cholesterol. The titration was carried out at a protein concentration of 6 μM in PBS at pH 7.4 and 37 °C. Cholesterol was added from an ethanol stock solution. The titration curve (●) follows a classical binding reaction pattern, with a 1:1 stoichiometry. In order to estimate the apparent KD, we have fitted the data with eqn (6) (solid line). The apparent KD* was 3×10−8. For comparison purposes, the binding curves assuming different KD* values and the Δ[θ]222 observed at saturation have been simulated with eqn (6) (broken lines). Values are means±S.D. of four independent experiments.

Figure 7 Proposed binding mechanism between StAR and cholesterol

(A) In the absence of cholesterol, the C-terminal α-helix of StAR can undergo a reversible folding–unfolding reaction. Cholesterol would bind to and dissociate from this partially unfolded state. Upon binding, cholesterol stabilizes the C-terminal α-helix. The position of Phe267 at the interface of the C-terminal α-helix and the cholesterol-binding site is indicated by the arrow. (B) The cmc and solubility of cholesterol as a monodisperse species (molecules in grey balls) are both in the 10−8 M range. Therefore, in an aqueous environment, monodisperse cholesterol is in equilibrium with micelles (depicted as the rod-shape oligomer). Consequently, based on the 1:1 stoichiometry, the formation of a StAR–cholesterol complex can proceed via desorption from the micelles (broken-lined arrows) and/or direct binding (solid arrows) to free cholesterol. In this context, the KD of 3×10−8 obtained from the binding curve (Figure 6, solid line) can be considered only as an apparent KD (KD*).


Currently, the mechanism of StAR activity is a matter of debate. There are indications that StAR could undergo a conformational change to a molten globular state upon interaction with the OMM where the pH is presumably more acidic than the cytoplasm. It has been suggested that this molten globule state of StAR could act as an on/off switch for cholesterol entry into the mitochondria [20]. Based on molecular modelling, we have proposed that, at physiological pH, StAR is in a dynamic equilibrium between a native well-folded form and a partially folded state where the C-terminal α-helix undergoes a folding/unfolding transition in the absence of cholesterol [22,39] (see also Figure 7A). The unfolding of the C-terminal helix is hypothesized to allow access of cholesterol to its binding site on StAR. Once loaded with one cholesterol molecule (Figure 7B), the refolding of the C-terminal helix would stabilize the complex. The unfolding of the C-terminal helix would then facilitate the transfer of cholesterol to the mitochondria either directly to the membrane or via a transporter [22]. Our model has recently received support by a molecular dynamics simulation showing that uptake and release of cholesterol by StAR/StART could occur via conformational changes localized at the vicinity of the binding site while preserving the native conformation of the rest of the protein [35]. The present study provides experimental evidence to support the model proposed in Figure 7.

In vitro activity of purified N-62 StAR

Given that a StAR–cholesterol association would be the first step in the cholesterol transport mechanism, we characterized the overall structure and cholesterol-binding properties of StAR. All experiments were carried out under physiological conditions since preferential binding of StAR to cholesterol was optimal at pH 7.4 (Supplementary Figure S2). We found that the wild-type protein displayed a higher activity than the α-helix-4-disrupted F267Q mutant (Figure 1). Furthermore, the addition of exogenous cholesterol was required to increase pregnenolone production by both the wild-type StAR and the F267Q mutant (Figure 1). On the other hand, the F267Q mutant significantly lost its binding capacity to cholesterol (Figure 2) as well as its steroidogenic activity (Figure 1), suggesting that α-helix 4 is involved in both the cholesterol binding and the steroidogenic activity of StAR.

These results indicate that a stable StAR–cholesterol interaction is a prerequisite for steroidogenic activity of mitochondria. This notion is supported by several StAR mutants that had an impairment in cholesterol binding which correlated with a loss of steroidogenic activity [21,40,41]. However, the clinical mutation R182L, which causes severe LCAH, was shown to bind cholesterol and transfer it between liposomes in vitro (presumably at room temperature). On the other hand, this mutant cannot generate a steroidogenic activity in mitochondria [42] and conflicts with the generally accepted cholesterol binding/steroidogenic activity relationship. Nonetheless, it is worth mentioning that the steroidogenic activity of StAR also relies on protein–protein interactions [43]. With this regard, we wish to emphasize that in our model [22] and that of Baker et al. [42], the Arg182 residue lies near the surface. Hence, it is possible that the R182L mutant has impaired protein–protein interactions, which could explain its lack of steroidogenic activity and the apparent conflict between cholesterol binding and the activity of this mutant. In addition, Baker et al. [42] have observed that the secondary-structure content (estimated from CD spectra) of the R182L mutant is similar to the wild-type (presumably at room temperature) [42]. However, similar secondary-structure contents do not mean that the tertiary structures are identical. Moreover, differences in tertiary structures could impair the ability of StAR R182L to interact with other proteins involved in cholesterol transfer into the mitochondria [43].

Effect of cholesterol on N-62 StAR overall structure and stability

In the absence of cholesterol, StAR had the characteristics of a well-folded protein. Indeed, the far-UV CD spectrum of StAR (Figure 3) was consistent with the secondary-structure content suggested by our model in the apo-state (Figure 7). Moreover the co-operative temperature-denaturation observed in Figure 4 clearly showed that the protein also possessed a stable tertiary structure. Although the F267Q mutant had a defined tertiary structure, a lower T° upon thermal denaturation indicated that the structure was less stable. This effect can be ascribed to the polarity of the replaced Glu267 residue which increases the exposure of buried hydrophobic residues of α-helix 4 to the solvent (Figure 7), thereby reducing the conformational stability of the protein.

The thermodynamic parameters characterizing the stability of the tertiary structure at 37 °C are listed in Table 1. As anticipated, and in accordance with our model (Figure 7), the addition of cholesterol stabilizes the α-helical content (Figure 4) and the tertiary structure (Figure 5) of the StAR protein. However, this is in marked contrast with a previous study by Petrescu et al. [41] reporting that cholesterol binding altered the secondary structure of StAR and dramatically reduced the proportion of α-helix. It should be emphasized that in the fluorometric study by Petrescu et al. [41], NBD (7-nitrobenz-2-oxa-1,3-diazole)–cholesterol was used and was shown to bind to StAR with a 2:1 stoichiometry [41], whereas cholesterol binds with a 1:1 ratio. According to modelling studies, only one molecule of cholesterol can fit into the hydrophobic pocket of StAR-related domains [21,22]. In contrast, NBD–cholesterol is much bulkier than cholesterol. In addition, the NBD moiety contains hydrogen bond acceptors and charged groups. This further lowers the likelihood that NBD–cholesterol may be found in the interior of the hydrophobic cavity of StAR. Hence, contrary to cholesterol, such physico-chemical properties will favour the formation of solvent-exposed interactions between NBD–cholesterol and StAR. This could explain the difference between the 1:1 stoichiometry determined in the present study and the apparent 2:1 stoichiometry previously reported [41]. With regards to the decrease in secondary structure upon addition of cholesterol, it is worth mentioning that we have observed a similar effect directly after the addition of cholesterol. However, as discussed, this decrease was only transient, as an increase in secondary structure (compared with the apo-form) occurred after the equilibrium was reached (i.e. approx. 90 min). On the other hand, such an equilibration step has not been reported in previous studies [41].

The contribution of cholesterol to the stabilization free energy of StAR [ΔΔGu° (37 °C)] is found to be of entropic origin. Indeed, the present results (Table 1) suggest that unfolding of cholesterol-bound StAR, compared with the apo state, is accompanied by a lower increase in entropy [T·ΔSu° (37 °C); 77.37±0.28 compared with 86.17±0.20 kcal·mol−1]. On the other hand, the enthalpy of unfolding of the complex is less positive than that of the apo-form. This property indicates that more favourable (enthalpic) interactions exist in the folded state of the apo-form. Compared with the hydrogen bonds, breaking hydrophobic interactions is associated with smaller ΔH and ΔS [4446]. Therefore the results of the present study suggest that hydrophobic interactions between the binding site of StAR and cholesterol stabilize the complex. In other words, it appears that specific hydrogen bonds are disrupted by the presence of cholesterol. This was anticipated considering that cholesterol is a highly hydrophobic molecule and that the binding site is composed of mainly hydrophobic residues with the exception of a putative salt bridge between residues Glu168 and Arg187 [22]. However, in F267Q where the hydrophobic interactions between α-helix 4 and the cholesterol-binding site were disrupted, the entropy of unfolding, even in the presence of cholesterol, was dramatically decreased. Given that ΔΔGu° of the mutant is only 0.30±0.16 kcal·mol−1 (Table 1), we conclude that StAR F267Q cannot bind to cholesterol to form a stable complex.

Cholesterol-induced helical modification of StAR at a physiological pH has allowed us to determine a 1:1 stoichiometry and a KD of 3×10−8 for the StAR–cholesterol complex, which is in agreement with those studies characterizing the cholesterol-binding site as a hydrophobic pocket which can accommodate only one cholesterol molecule [21,22,35].

Considering that the maximum solubility of cholesterol in an aqueous solution is 5 μM with monomeric micellar concentrations and cmcs of 1×10−8 M and 4×10−8 M respectively [38], a question arises as to whether StAR binds to monomeric cholesterol and/or desorbs it from micelles. Based on the KD value, it is possible to state that monomeric cholesterol can either bind to StAR or associate with micelles. A comparison of the free energy of micellization [47] which is given by −RT·ln cmc (∼−10 kcal·mol−1) with that of the StAR–cholesterol complex formation (−RT·ln KD∼−12.5 kcal·mol−1) suggests that binding of cholesterol to StAR is favoured over micelles. Although self-association of cholesterol might be cumbersome for the determination of a ‘real’ KD, it appears that this phenomenon mimics, to some extent, what occurs in vivo. In fact, in the cytoplasm, cholesterol partitions into membrane reservoirs and droplets [48]. Nevertheless, StAR binds cholesterol in vivo and transfers it to the mitochondria. Whether the formation of the StAR–cholesterol complex proceeds via the binding of monomeric cholesterol and/or desorption from membranes or droplets is not clear. However, these results suggest that both are possible.


This work was supported by a grant (MT-10983) to J.-G.L. from the Canadian Institute of Health Research. We would like to thank Ms Andrée Lefebvre for her excellent technical assistance and Dr Van Luu-The (Oncology and Molecular Endocrinology Research Center, Laval University, Laval, Quebec, Canada) for the gift of epostane. We are also grateful to Dr Gilles Dupuis for critically reading this manuscript.

Abbreviations: cmc, critical micellar concentration; IMM, inner mitochondrial membrane; LCAH, lipoid congenital adrenal hyperplasia; NBD, 7-nitrobenz-2-oxa-1,3-diazole; OMM, outer mitochondrial membrane; StAR, steroidogenic acute regulatory protein; StART, StAR-related lipid transfer; TCEP, tris-(2-carboxyethyl)phosphine


View Abstract