Steroidogenesis depends on the delivery of free cholesterol to the inner mitochondrial membrane by StAR (steroidogenic acute regulatory protein). Mutations in the StAR gene leads to proteins with limited cholesterol-binding capacity. This gives rise to the accumulation of cytoplasmic cholesterol, a deficit in steroid hormone production and to the medical condition of lipoid congenital adrenal hyperplasia. A detailed understanding of the mechanism of the specific binding of free cholesterol by StAR would be a critical asset in understanding the molecular origin of this disease. Previous studies have led to the proposal that the C-terminal α-helix 4 of StAR was undergoing a folding/unfolding transition. This transition is thought to gate the cholesterol-binding site. Moreover, a conserved salt bridge (Glu169–Arg188) in the cholesterol-binding site is also proposed to be critical to the binding process. Interestingly, some of the documented clinical mutations occur at this salt bridge (E169G, E169K and R188C) and in the C-terminal α-helix 4 (L275P). In the present study, using rationalized mutagenesis, activity assays, CD, thermodynamic studies and molecular modelling, we characterized the α-helix 4 mutations L271N and L275P, as well as the salt bridge double mutant E169M/R188M. The results provide experimental validation for the gating mechanism of the cholesterol-binding site by the C-terminal α-helix and the importance of the salt bridge in the binding mechanism. Altogether, our results offer a molecular framework for understanding the impact of clinical mutations on the reduction of the binding affinity of StAR for free cholesterol.
In steroidogenic cells, cholesterol is converted into pregnenolone by the cytochrome P450 side-chain-cleavage enzyme which is located on the matrix side of the IMM (inner mitochondrial membrane) [1,2]. StAR (steroidogenic acute regulatory protein) (STARD1) regulates cholesterol transfer from cellular pools via the OMM (outer mitochondrial membrane) to the IMM [3–5]. This event represents the rate-limiting step in the production of steroid hormones [1,2,5,6]. StAR was first cloned by Clark et al. , who identified it as being the same protein originally described by Orme-Johnson and colleagues [7–9], a phosphoprotein which is responsible for adrenocorticotropin-hormone-stimulated steroidogenesis. Under the influence of tropic hormones [9,10], StAR is acutely expressed as a 285-amino-acid proprotein in the cytoplasm with an apparent molecular mass of 37 kDa, and StAR-induced steroidogenenis is regulated by phosphorylation [11–14]. Upon translocation into mitochondria, StAR is proteolytically cleaved to yield a 30 kDa protein ; however, it was established that StAR acts outside of mitochondria . StAR is also found to interact with proteins located at the OMM [17,18]. It is believed that the interaction of StAR–cholesterol with these OMM proteins leads to the delivery of free cholesterol to an import complex. This illustrates a multifaceted mechanism of StAR-mediated mitochondrial cholesterol import.
StAR possesses two distinct domains, an N-terminal mitochondrial-import sequence [16,19], and a START (StAR-related lipid transfer) domain responsible for the binding and transfer of cholesterol [20,21]. START domains are 200–210-amino-acid lipid/sterol binding motifs that are conserved in animals, plants, bacteria and unicellular protists . They are implicated in a variety of biological functions, such as lipid transport, metabolism, signal transduction and transcriptional regulation [23,24]. It was proposed that all START domain-containing proteins include a binding pocket, where modifications in that pocket would determine ligand-binding specificity and function [25,26]. The high affinity of these proteins for their ligands, including cholesterol, phosphatidylcholine, phospholipids and sphingolipids , suggests a ligand-binding mechanism common to the START domain superfamily . Investigation of the structure of these proteins, and the mechanism and biological relevance of their lipid-binding activity, has been the subject of intense research in recent years.
Among the fifteen START domain proteins identified so far , the crystal structure of the START domain of human MLN64 (metastatic lymph node 64)/StARD3 was the first to be resolved . On the basis of the sequence identity between the START domains of StAR and MLN64 and their functional similarities in binding and transferring cholesterol , three-dimensional models of StAR in the apo and cholesterol-bound states were generated . Subsequently, two additional, but virtually identical, models were proposed [30,31]. Not surprisingly, our model of the apo form revealed the presence of a cavity large enough to fit one cholesterol molecule. This cavity was proposed to promote the local unfolding of the C-terminal α-helix 4 to allow for the binding of cholesterol. We recently validated experimentally this local unfolding or opening of StAR and the existence of the partially unfolded state .
Thirty mutations in the StAR gene are known to cause LCAH (lipoid congenital adrenal hyperplasia), a severe autosomal and recessive form of congenital adrenal hyperplasia [33,34]. These mutations lead to the production of non-functional StAR proteins. Many of these mutations yield gene products that lack the ability to bind to free cholesterol. This leads to the accumulation of cholesterol in the cytoplasm and an impaired steroid synthesis in LCAH patients. According to the molecular models of StAR, these mutations are located either at the C-terminal α-helix 4 or at the cholesterol-binding site [29,34]. For example, the clinical mutation of Glu169 to glycine or lysine causes LCAH . Furthermore, the E169L or R188M mutations showed a dramatic loss in the steroidogenic activity in transfected COS-1 cells . Therefore such mutations confirm the importance of preserving the integrity of the putative salt bridge in StAR. Other clinical mutations affect the C-terminus of StAR, in particular α-helix 4. For instance, the mutation of Leu275 to proline causes LCAH and exhibits only 13% of steroidogenesis activity compared with the WT (wild-type) protein, as determined in COS-1 cells . Moreover, a truncation of ten C-terminal residues reduced the activity of StAR by 70%, whereas a deletion of 28 residues abolished all activity in transfected cells . A salt bridge (Asp332–Arg351) is present in MLN64 and was proposed to be important for the specific interaction with the 3β-hydroxyl group of cholesterol . The molecular modelling of StAR revealed this key molecular determinant between Glu169 and Arg188 at the bottom of the cavity [29–31]. Consequently, we hypothesized that the salt bridge was critical for specific cholesterol binding. Taken together, these observations support the predictions that the C-terminal α-helix 4 and the Glu169–Arg188 salt bridge play critical roles in the binding process.
Basic knowledge, such as the three-dimensional structure of StAR (both in the apo and holo forms), as well as the mechanism of the reversible and specific binding of free cholesterol, is still lacking. In the present study, we focus on the experimental validation of our three-dimensional molecular model of StAR  and on the mechanism of free cholesterol binding by the protein. Moreover, we sought to investigate, in more detail, the role of Glu169–Arg188 salt bridge in StAR and α-helix 4 in the binding mechanism of free cholesterol using structure-based mutagenesis, binding and activity assays and biophysical methods. Our results show that weakening of the specific hydrophobic tertiary interactions involving the C-terminal α-helix 4 reduces the secondary structure of StAR and the apparent binding affinity above the solubility limit of free cholesterol. As discussed, this is expected to cause a reduction in the cholesterol import and the steroidogenic activity. More precisely, we observe that the clinical mutation L275P retains only a fraction of the apparent binding affinity and the steroidogenic activity of WT StAR. In addition, the double mutation E169M/R188M significantly stabilizes the apo form of StAR and abolishes both the binding and steroidogenic activity. These results are in accordance with a reduction in the population of the partially folded state, which is suggested to initiate the specific binding of free cholesterol. Altogether, our results constitute a further validation of our three-dimensional molecular model of StAR and its mechanism of specific binding of free cholesterol. As discussed, our results also provide a framework to understand the origin of LCAH-promoting mutations in the StAR gene.
All restriction and modifying enzymes were purchased from New England Biolabs. Oligonucleotides were purchased from Invitrogen. The cloning vector pET-21b was purchased from Novagen. BL21 (DE3) Escherichia coli competent cells were obtained from Stratagene. [1,1,6,7-3H]Pregnenolone (95.0 Ci/mmol) was purchased from Amersham Biosciences. (22R)-Hydroxycholesterol and cholesterol were purchased from Sigma–Aldrich. Nickel affinity resin was from Qiagen. Antisera against pregnenolone was obtained from ICN Biochemicals. All other chemicals were purchased from local suppliers.
For the salt bridge double mutation, two series of PCR amplifications were performed using the N-terminal His6 (hexahistidine)-tagged N-62 StAR (construct lacking the N-terminal 62 residues; StAR WT) as a DNA template . The E169M mutant was constructed and used as a template for the subsequent R188M mutation, thereby generating E169M/R188M. The α-helix 4 mutations were performed using N-62 StAR to create L271N and L275P. All constructs were cloned into the pET-21b expression vector using the NheI and BamHI restriction sites and the following primers: 5′-ATTCTAGCTAGCCACCACCACCACCACCACCTGGAAGAGACTCTCTACAGTGACC-3′ and 5′-CGGATCCTCAACACCTGGCTTCAGAGGC-3′.
Protein expression and purification
The proteins were expressed and purified as described previously , and were stored at room temperature (25°C). The protein concentration was determined by measuring the absorbance at 280 nm, considering a molar extinction coefficient of 27080 M−1·cm−1 [calculated using VectorNTI software (Invitrogen)].
Steroidogenic activity assay
Mitochondria were isolated from MA-10 cells as described previously [32,37] and were freshly used or stored at −80°C for later experiments. Mitochondria (1 μg/μl) and purified StAR proteins (10 μM) were subjected to an In vitro assay system  with a modified buffer containing 5 mM sodium succinate and 1 μM 3β-hydroxysteroid dehydrogenase inhibitor (epostane). The amount of pregnenolone produced was quantified by RIA . The activity was normalized according to the following equation (eqn 1):
where A is the activity, Amin is the minimal activity (buffer control) and Amax is the maximal activity (StAR WT). Thus StAR WT activity was set to 100% and the buffer control was 0%.
StAR radioligand-binding assay
Binding assays were performed using [3H]cholesterol as a radiolabelled ligand. Protein (2.5 μg/ml) was incubated with 1 pmol [3H]cholesterol in the presence or absence of 5 μM unlabelled cholesterol for 60 min at room temperature in 100 μl of PBS [50 mM NaH2PO4 and 150 mM NaCl (pH 7.4)]. After 60 min incubation, the total volume was increased to 500 μl with 100 mM PBS containing 0.3% gelatin. To distinguish the effect of dilution on StAR–cholesterol dissociation, some binding experiments were performed directly in the total volume of the standard buffer. Bound and free cholesterol were separated as described previously . Prior to centrifugation of the RIA mixture, 0.01% Triton X-100 was added to reduce the binding of StAR to the spin tubes and to avoid local aggregation. Radioactivity of bound [3H]cholesterol was measured with a Beckman LS8000 CE Scintillation Counter (Beckman Coulter). Background levels were subtracted from the total counts of the samples. The relative binding was calculated as follows (eqn 2):
where B is the bound ligand, Bmin is the control not containing the protein and Bmax is the maximal binding (StAR WT). StAR WT cholesterol binding was set to 100% and the controls were set to 0%. The relative displacement was calculated according to the following equation (eqn 3):
where D is the displaced ligand and B is the bound ligand. Statistical analysis was performed using SigmaStat software (Systat Software).
CD measurements were carried out as described previously  using a protein concentration of 0.16 mg/ml in 25 mM phosphate buffer (pH 7.4). Cholesterol-induced changes in the secondary structure were evaluated by monitoring the ellipticity in the far-UV region (190–250 nm). Thermal denaturation curves were established for StAR in the absence or presence of cholesterol at pH 7.4 by monitoring the CD values at 222 nm while increasing the temperature from 10°C to 90°C. In the experiments including cholesterol, StAR was pre-incubated for 45 min with the substrate. The temperature-induced denaturation curves monitored by CD were acquired as described previously . The temperature dependence of the mean residue ellipticity at 222 nm (θ222) was fitted using an in-house non-linear least-squares fitting program . The α-helix content was estimated with the software DICROPROT  by fitting the data using the molar ellipticity method at 220 nm .
The thermodynamical parameters45,46]. The observed value of θ at any temperature is given by θ=θfPu+θuPu, where θf and θu are the values of θ in the folded and unfolded states respectively and Pu and Pf are the population unfolded and folded values respectively. In this approach , the temperature dependence of the molar ellipticity (θ) is given by equation (4):
where θn(0) and θu(0) are the mean residue ellipticities at 0°C for the folded and unfolded states respectively and dθn(T)/dT and dθu(T)/dT are the constant slopes of θn(T). The population of the unfolded state is given as follows (eqn 7):
where R is the gas constant and T is the absolute temperature. Therefore in order to get Pu(T),42] by the determination of T°, the melting temperature and (eqn 8):
In the present study, the heat capacity of unfolding ΔCp,u was considered to be constant and equal to 1 kcal·mol−1·K−1 (1 cal≈4.184 J).
All of the modelling was performed with the Insight II 2000 suite (Accelrys Software) running on a Silicon Graphics Octane 2 computer. More specifically, all of the mutations were performed on the three-dimensional model of StAR WT (PDB ID: 1img) using the Biopolymer module. The minimization of the potential energy of the different mutants was performed using the Discovery module and the CVFF force-field using the conjugate gradient algorithm and turning off all coulombic interactions. All the rendering was performed using PyMOL (DeLano Scientific).
Structure-based mutagenesis of StAR
To assess the role of the C-terminal α-helix 4 during cholesterol binding, we generated the L271N and the L275P StAR mutants. The reasoning behind these mutations is illustrated in Figure 1. The StAR model reveals the existence of a hydrophobic core or interface involving the C-terminal α-helix 4 where the Leu271 and Leu275 side chains are deeply buried. Also, another hydrophobic cluster involving Met225, Phe267 and Trp241 is present (Figure 1A). The modelling of the L275P mutant suggests that the cyclization of the Pro275 side chain will reduce the extent of burial of the hydrophobic surface and weaken the strength of this hydrophobic interface (Figure 1B). In order to further verify the importance of this interface, we mutated Leu271 to an asparagine residue, resulting in the replacement of the leucine side chain with the isosteric polar asparagine side chain (Figure 1C). The unfavourable desolvation of the asparagine side chain upon its burial at the hydrophobic interface can be expected to favour the local unfolding of α-helix 4.
Molecular model of the StAR–cholesterol complex
Furthermore, to study the role of the conserved Glu169–Arg188 salt bridge, we generated the E169M/R188M double mutant. The rationale for such a double mutant was to replace the ionic and hydrogen-bonding interactions between the carboxylate group of Glu169 and the guanidino moiety of Arg188 with stable hydro- phobic interactions (Figure 2). It was reasoned that such a mutant would preserve a stable tertiary structure and allow for the assessment of the role of the polar groups in the binding of free cholesterol with minimal perturbation. In accordance with this assertion, note that the volume of the two methionine side chains is very close to the sum of the volumes of a arginine and a glutamic acid side chain.
Model of the StAR salt bridge mutant
In vitro steroidogenic activity is affected in the StAR mutants
The steroidogenic activity of the three StAR mutants was assessed In vitro using mitochondria isolated from steroidogenic Leydig MA-10 cells. This assay measured the production of pregnenolone from endogenous cholesterol present in the OMM or from free cholesterol exogenously added into mitochondria. Incubation of StAR WT and cholesterol with mitochondria increased the level of pregnenolone production to a level comparable with the (22R)-hydroxycholesterol control (Figure 3). (22R)-Hydroxycholesterol crosses the OMM freely and can be transformed into pregnenolone readily without the need for StAR. In contrast, the activity of the L271N and L275P mutants was decreased to 29±5% and 13±2% respectively. As for the E169M/R188M mutant, the loss in the cholesterol-transfer capacity was more pronounced, with only 2±1% of the pregnenolone production of StAR WT (Figure 3 and Table 1). These results indicate that the L275P and L271N mutations at the hydrophobic interface (Figures 1B and 1C) impaired the steroidogenic activity of StAR. This is consistent with a previous study, where the F267Q mutation yielded only 23% of the activity of StAR WT . Phe267 is expected to interact directly with cholesterol, so we suggested that the F267Q mutation would weaken the stability of the StAR–cholesterol complex. Furthermore, our results corroborate with the truncation experiments in α-helix 4, where a 10-amino-acid deletion from the C-terminus reduced the activity of the protein by 70% .
Steroidogenic activity of StAR WT and the StAR mutants
|.||Relative .||Relative .||Relative .|
|Protein .||activity (%) .||binding (%) .||displacement (%) .|
|.||Relative .||Relative .||Relative .|
|Protein .||activity (%) .||binding (%) .||displacement (%) .|
Cholesterol binding is impaired in StAR mutants
Classical binding assays were performed to assess the apparent and relative cholesterol-binding affinity of the different mutants in comparison with StAR WT (Figure 4). As demonstrated by the radioactive counts, StAR WT bound radiolabelled cholesterol with 20-fold greater affinity than the controls. The addition of unlabelled cholesterol caused 55±4% displacement of labelled cholesterol, confirming the binding specificity. However, the apparent and relative binding affinities of StAR L271N (42±4%) and StAR L275P (13±3%) mutants were observed to be significantly decreased compared with StAR WT. Interestingly, the E169M/R188M mutant displayed non-significant relative binding (4±3%), suggesting a loss in the ability of this mutant to bind cholesterol (Figure 4 and Table 1). These results are consistent and commensurate with the decreased potential of the different mutants to promote steroidogenesis in mitochondria (Figure 3).
Competitive binding of cholesterol to StAR WT and the StAR mutants
Structural and thermodynamical chraracterization of the StAR mutants
To investigate whether the loss of biological function is a result of structural modifications, we proceeded to characterize the overall structure of the mutants by CD spectroscopy. We show in Figure 5 the far-UV CD spectra of StAR WT and its mutants. In the absence of cholesterol, StAR WT has a 32.5% α-helical content, which is close to the 40% maximum helical content calculated for the three-dimensional model. The difference between the actual and the maximum helical content correlates with the existence of a significant population of partially folded states of the apo protein. The L271N and L275P mutants showed a reduced α-helical content of 28.8% and 29.4% respectively compared with StAR WT (Figure 5 and Table 2). As discussed above, these mutants were expected to lead to the weakening of the hydrophobic interface and hence to a reduction in the amount of α-helical content and stable tertiary structure. On the other hand, the E169M/E188M mutant had a remarkable increase in α-helical content up to 40.9% compared with StAR WT. This is suggestive of the stabilization of the secondary and tertiary structure in StAR. As demonstrated previously, the addition of cholesterol to StAR WT resulted in an increase in the actual helical content by 6.1% . This is the result of the binding and the stabilization of StAR in its fully folded state. However, the addition of cholesterol had only a marginal effect on all of the mutants, with variations of ±2% in their α-helical content (Table 2). These results are in agreement with the decrease and loss of binding/activity of the StAR mutants (Figures 3 and 4).
Effect of cholesterol binding on the secondary structure of StAR mutants
|.||.||.||.||Melting temperature .|
|.||α-Helix content (%) .||T° (°C) .|
|Protein .||−Ch .||+Ch .||Δ .||−Ch .||+Ch .||Δ .|
|.||.||.||.||Melting temperature .|
|.||α-Helix content (%) .||T° (°C) .|
|Protein .||−Ch .||+Ch .||Δ .||−Ch .||+Ch .||Δ .|
To further characterize the mutants, we performed thermodynamic analyses based on the thermal denaturation profiles. The unfolding profiles of the StAR mutants all showed single co-operative transitions, indicating a well-defined tertiary structure, while following a two-step process of denaturation (Figure 6). From these plots, we derived the corresponding stability curves (Figure 7), the melting temperature T° (Table 2) and the thermodynamic parameters at 37°C (Table 3). In the absence of cholesterol, StAR L271N and L275P had a T° of 40.2°C and 38.2°C respectively, a decrease compared with the WT protein, which had a T° of 42.4°C. Interestingly, the E169M/R188M mutant showed an important increase in its T° to 50.1°C. Despite the T° variations in the mutants, the addition of cholesterol caused only a slight increase in the melting point by 1.0–1.4°C, whereas StAR WT had a 3.6°C increase in melting point on addition of cholesterol (Table 2). The higher T° indicates a stabilization of the holo protein. As reported for StAR WT previously , the free energy of unfolding of the holo state had a higher valueTable 3). This increase in the free energy of unfolding indicates a favourable interaction between StAR and cholesterol. As a consequence, an entropy-driven stabilizing effect occurred, since TTable 3). This indicates that the presence of cholesterol provided little stabilization of the mutants, possibly as a result of reduced binding (Figure 4).
Thermal denaturation of StAR WT and its mutants
Stability curve of the StAR mutants
|Protein .||−Ch .||+Ch .||−Ch .||+Ch .||−Ch .||+Ch .||(kcal·mol−1) .||(kcal·mol−1) .||(kcal·mol−1) .|
|Protein .||−Ch .||+Ch .||−Ch .||+Ch .||−Ch .||+Ch .||(kcal·mol−1) .||(kcal·mol−1) .||(kcal·mol−1) .|
Molecular modelling of the salt bridge mutant
Although the outcome of the L271N and L275P mutations on the binding, steroidogenic activity, secondary and tertiary structures of StAR can be rationalized, the result of the E169M/R188M salt bridge mutation is more puzzling. Indeed, although the loss of binding could be explained by the removal of the conserved salt bridge and hence the possibility of forming specific interactions, the double mutant is more helical and thermodynamically stable. It is possible that the mutation led to the stabilization of the fully folded state and abolished the population of the partially folded state that binds cholesterol (Figure 8). In order to explore this possibility, we modelled the putative structural changes induced by the E169M/R188M mutation by simple potential energy minimization. As depicted in Figure 2, the model of the E169M/R188M mutant shows the establishment of new tertiary and hydrophobic interactions between Met141, Met144, Met225, Trp241 and Phe267, in addition to a reduction in the size of the binding site. Compared with StAR WT, the binding site in the E169M/R188M mutant is too small to accommodate cholesterol. Therefore it is reasonable to suggest that the E169M/R188M mutation stabilizes the closed state through hydrophobic interactions. Added to the absence of polar group, we propose that these effects are responsible for the lack of binding.
Mechanism of reversible cholesterol binding by StAR
As research into the mechanism of action of StAR progresses, a clearer understanding of its structure and ligand-binding process becomes one of the focal points in the elucidation of its mode of action. In the present study, we focused on two domains which are necessary for StAR's activity, the cholesterol-gating domain, represented by the C-terminal α-helix 4, and the cholesterol-binding site buried deeply within the protein's structure.
Many lines of evidence support the fact that StAR must undergo a conformational change in order to bind free cholesterol [13,29–31]. More precisely, it is believed that the C-terminal α-helix 4 undergoes a reversible unfolding/folding transition to allow cholesterol to access its binding site and the subsequent stabilization of the complex . Mutating Leu271 to an asparagine residue at α-helix 4 led to a dramatic decrease in the binding and activity of StAR. Similar results were obtained with the L275P mutant. Furthermore, COS-1 cells transfected with the StAR L275P construct showed 13% activity in comparison with StAR WT . This is in agreement with our In vitro mitochondrial assay (Figure 3), as well as with Bose et al. , who reported that the L275P mutant converted 10% more cholesterol into pregnenolone than the control vector . Also, the mutations in α-helix 4 caused a reduction in the α-helical content and in the melting temperature of StAR (Table 2). However, the α-helix 4 StAR mutants bearing these new properties did not exhibit any secondary structural changes nor an improvement in T° in the presence of cholesterol, as noted for StAR WT.
StAR clinical mutations that lead to gene products with reduced cholesterol-binding capacity have been described previously [33–35]. The best-characterized clinical mutations that affect cholesterol binding are M225T, L275P, E169G and E169K . It is interesting to note that Met225 and Leu275 are part of a hydrophobic core involving the C-terminal α-helix 4 (Figure 1A). Hence we proposed that weakening the hydrophobic core should destabilize the fully folded state of the C-terminal α-helix 4 both in the free and bound states of StAR. As shown here, mutants that reduce the extent of this hydrophobic core (L271N and L275P) diminished the helical content of StAR both in the apo and holo forms. These mutants also displayed lower steroidogenic activities and lesser apparent binding affinities compared with StAR WT. In a previous report, we determined that StAR WT binds cholesterol with an apparent Kd of ∼8 × 10−8 . It is also interesting to note that this apparent Kd is of the same order of magnitude as the critical micellar concentration and the solubility limit of monodispersed free cholesterol. We attempted to measure the apparent Kd of the different mutants described in the present study, however we could not obtain accurate values because of a lack of significant specific cholesterol binding (Table 1), and a lack of protein stabilization or structural changes induced by the ligand (Table 2). It is therefore clear that the L271N and L275P mutations in StAR decrease the apparent affinity for cholesterol. It can be assumed that their actual Kd will become larger than the actual concentration of free cytoplasmic cholesterol. The end result would be that such mutants are expected to lead to the accumulation of free cholesterol in the cytoplasm, the decrease in cholesterol import into mitochondria and eventually to LCAH.
According to our three-dimensional model of StAR, Glu169 is part of a conserved salt bridge with Arg188, which may be involved in a specific interaction with cholesterol (Figure 1A and Figure 2A). Interestingly, a salt bridge is present in MLN64 between Asp332 and Arg351, where it is observed to form hydrogen bonds with the 3β-hydroxyl group of cholesterol . A few studies have addressed the salt bridge of StAR. For instance, in transfected COS-1 cells, the E169M and R188M mutations of the salt bridge individually reduced the activity of StAR , and the newly described R188C clinical mutant, which causes non-classic LCAH, showed a reduced binding of cholesterol and steroidogenic activity .
In the present study, we generated the E169M/R188M double mutant to test whether it can completely abolish cholesterol recognition. Indeed, this double mutant had lost all of its cholesterol-binding capacity, and consequently proved to be inactive steroidogenically. However, in constrast with the clinical mutants, the E169M/R188M mutant, when compared with StAR WT, displayed a higher α-helical content and thermodynamical stability; these augmentations were rationalized on the basis of the putative formation of favourable hydrophobic interactions that would stabilize the C-terminal α-helix 4 in its closed conformation (Figure 8), and consequently this would explain the lack of cholesterol binding. In addition to providing specific interactions with cholesterol, our results indicate that the salt bridge also has a role in the unfolding of the C-terminal α-helix 4 by promoting the opening of the binding site. In regards to the clinical mutations E169G, E169K  and R188C , it is clear that they will alter the conformational stability of the salt bridge and thereby change the shape and volume of the binding site. This is expected to impede on the favourable binding of cholesterol.
These results further support the notion that a stable StAR–cholesterol interaction is a pre-requisite for the steroidogenic activity of mitochondria . Indeed, several StAR mutants that had impaired cholesterol binding correlated with a loss of steroidogenic activity [20,21,48]. However, the R182L clinical mutation, which causes severe LCAH, can bind cholesterol and transfer it between liposomes In vitro, but has no steroidogenic activity in mitochondria . This conflicts with the generally accepted ligand binding/protein activity relationship. Also, it is worth mentioning that the steroidogenic activity of StAR also relies on protein–protein interactions . Accordingly, we wish to emphasize that in our model  and that of Baker et al. , the Arg182 residue is located near the surface of StAR. In addition, it is now evident that StAR must interact with mitochondrial proteins, such as TSPO (translocator protein), to deliver cholesterol . Therefore the R182L mutation may cause disruptions in these protein–protein interactions without affecting cholesterol binding. This explains its absent steroidogenic activity and the noticeable conflict between cholesterol binding and the activity of this mutant. In addition, according to their CD estimations, Baker et al.  observed that the secondary structure content of the R182L mutant is similar to StAR WT. However, comparable secondary structure contents do not imply that the tertiary structures are identical. Therefore, differences in tertiary structures, in particular the loop where Arg182 lies, may be the factor affecting the ability of StAR R182L to interact with other proteins involved in cholesterol transfer into the mitochodria .
The data, modelling and mechanisms in the present study offer a detailed description of the specific binding of free cholesterol by StAR. However, it is important to bear in mind that cholesterol binding and the formation of a stable StAR–cholesterol complex is only the initial step in steroidogenesis. Indeed, the cholesterol bound by StAR must be delivered to the IMM to be enzymatically transformed. The detailed mechanism of the delivery process is under study and is proposed to involve phosphorylation events [50,51], as well as interactions with mitochondrial membrane proteins and/or a delivery complex . Although our findings do not inform us directly about these processes, phosphorylation of StAR and/or its interaction with membrane proteins must stabilize the open state of the complex to favour the release and delivery of cholesterol to mitochondria (Figure 8). StAR may also be recycled before its degradation, given the high transport activity of the newly synthesized protein (400 molecules of cholesterol trafficked per StAR molecule per min) . Hence it will be important to study these mechanisms further in order to gain a complete understanding of the delivery process.
In conclusion, the results of the present study strengthen the notion that α-helix 4 of StAR acts as a gate that allows free cholesterol access to its binding site. Also, we provide strong evidence that key hydrophobic residues in α-helix 4 are involved in the fine-tuning of this gating action. Indeed, mutation of residues that reduce the effective hydrophobic surface of their side chains destabilize the fully folded state of the apo and holo forms of StAR. These mutations lead to StAR proteins with limited steroidogenic activities and lower apparent binding affinities. Because the apparent affinity of StAR WT for free and monodispersed cholesterol is in the same range as its concentration (e.g. ∼10−8 M), these mutations are expected to promote the accumulation of cholesterol in the cytoplasm. Most importantly, the locations of these residues on the primary structure corresponds to known clinical mutations on StAR that lead to cholesterol accumulation and ultimately to LCAH. Therefore our results provide a mechanistic framework to understand the molecular origin of the role of mutations in the StAR gene that cause LCAH.
We thank Ms Andrée Lefebvre for her excellent technical assistance and Dr Van Luu-The (Oncology and Molecular Endocrinology Research Center, Laval University, Québec, Canada) for the gift of epostane.
This work was supported by the Canadian Institute of Health Research [grant number MT-10983] (to J.-G.L).
These authors contributed equally to this work.