The steroid hormones — which are required for the homeostatic maintenance of blood pressure, carbohydrate metabolism and reproductive function — represent some of the most important molecules found in the body. Their synthesis is regulated by signals from the anterior pituitary gland that act on specific steroidogenic cells found mainly in the adrenal glands and gonads. In response to these signals the steroidogenic acute regulatory protein (StAR) is synthesized. This protein is required for the rapid increase in steroid hormone production and mediates the delivery of cholesterol to the enzyme — cytochrome P450 side chain cleavage (P450 scc) enzyme — that converts cholesterol to pregnenolone, the first steroid formed1 in the steroidogenic pathway. This enzyme resides in the inner mitochondrial membrane, and it has long been known that the transfer of cholesterol from the outer mitochondrial membrane to this enzyme is the rate-limiting step in steroidogenesis2.

To date, the details of how the biosynthesis of steroid hormones is regulated have remained obscure. Now, the paper by Tsujishita and Hurley3 in the May issue of Nature Structural Biology reports a high resolution tertiary structure of a critical region, called the START domain (for StAR related lipid transfer domains), of the MLN64 protein, a 50 kDa protein of unknown function that is highly and specifically expressed in the malignant cells of breast carcinomas. This START domain contains a hydrophobic tunnel that appears capable of binding a single cholesterol molecule. Thus, this finding is a first step toward understanding cholesterol transfer.

START domains are found in a wide variety of proteins including the phosphatidylcholine transfer protein, acyl-CoA thioesterase, p122-RhoGAP, the Goodpasture antigen binding protein and, most importantly for the work described here, MLN64 and StAR4. They are ∼200-residue lipid-binding motifs4, and the importance of the new structure is that the START domain in MNL64 is highly homologous to the START domain in the StAR protein that is required for the regulation of steroid biosynthesis.

Many studies have demonstrated excellent correlations between StAR expression and steroid hormone biosynthesis. For example, expression of StAR through transfection results in increases in the transfer of cholesterol to the inner mitochondrial membrane and in steroid biosynthesis by both steroidogenic and nonsteroidogenic cells5,6,7. Therefore it is clear that StAR can somehow mediate cholesterol transfer to the inner mitochondrial membrane. Following the successful cloning of StAR in 1994 (ref. 8), one of the most important undertakings in this area became a quest for understanding how StAR mediates cholesterol transfer.

Early models (Fig. 1) speculated that StAR promotes cholesterol transfer by forming contact sites between the outer and inner mitochondrial membranes during the course of its import into the mitochondria5. However, this model required alteration when it was demonstrated that N-terminal truncations of the StAR protein, which removed as many as 62 amino acids and presumably could not be imported into mitochondria, supported full steroid production when transfected into COS-1 cells9 or when analyzed in an in vitro system with isolated mitochondria10.

Figure 1
figure 1

Previously proposed models for StAR function.

Full size image

In contrast, expression of StAR modified by removal of the C-terminal 28 amino acids resulted in a complete loss of steroid production9,11. Importantly, this is the location of the START domain in the protein. The importance of the C-terminal region in cholesterol transfer is suggested by the observation that MLN64 , a protein that may be involved in cholesterol transfer in breast cancer tumors, has significant homology to the C-terminal region of StAR and can promote cholesterol transfer12, as well as by the findings that virtually all of the mutations causing lipoid cogenital adrenal hyperplasia (CAH) are found in the C-terminus of StAR13. Nevertheless, the exact role of the START domains of StAR and MLN64 in promoting cholesterol transfer has been unclear.

Recent studies have demonstrated that StAR can act as a sterol transfer protein and can enhance sterol desorption from one membrane to another14. In this model, StAR is directed to the mitochondria via its N-terminus and then, utilizing C-terminal sequences, produces as yet unidentified alterations in the outer mitochondrial membrane that result in the transfer of cholesterol from the outer to the inner membrane. These studies suggested that transfer is specific for cholesterol; experiments employing phosphatidylcholine failed to show transfer of this phospholipid. Such desorption of cholesterol from the sterol-rich outer membrane to the sterol-poor inner membrane would serve to enhance steroid synthesis, explaining the requirement for StAR in this process15.

Investigations of StAR structure-function relationships in cholesterol transfer have been conducted, to try to explain the exact role of StAR. Miller and colleagues16 subjected StAR to limited proteolysis at different pH values and found that the molecule is altered as the pH decreases. They showed that StAR can form a molten globule in the pH 3.5–4.0 range and proposed that if the mitochondrial microenvironment is acidic, the StAR molecule may undergo a conformational shift. They proposed that this shift could form an extended structure and increase the flexibility of the linker region located between the N-terminus and the biologically active C-terminus. They further hypothesized that as the transition to a molten globule occurred, this structural change would lower the energy required to open the StAR structure, possibly exposing a cholesterol channel and/or lengthening the time that StAR resides on the outer membrane, allowing increased transfer of cholesterol during this period (Fig. 2). While each of the models mentioned above offers features that explain the apparent activity of StAR, new information is required to distinguish these models and provide new insights into how StAR mediates cholesterol transfer.

Figure 2
figure 2

New possibilities for StAR function.

Full size image

One of the most sought after pieces of information in the study of StAR has been a high resolution structure — in particular, a structure of the cholesterol transferring START domain. However, purifying and crystallizing the StAR protein, or truncated versions, proved problematic. Thus, Tsujishita and Hurley3 decided to focus instead on the START domain of MLN64, which is highly homologous to the StAR-START domain and was easier to analyze in vitro.

It was important that Tsujishita and Hurley3 were able to show that MLN64-START is indeed functionally similar to StAR-START. They demonstrated that both StAR-START and MLN64-START could bind cholesterol and that the binding occurred in a ratio of 1:1. They also demonstrated that neither START domain could bind cholesteryl oleate, suggesting similar specificities and steric properties. Thus, it follows that structural information about MLN64-START should be helpful for understanding the function of StAR-START.

The crystal structure of MLN64-START at 2.2 Å shows an α+β fold built around a U-shaped incomplete β-barrel. Most importantly, the crystal structure of MLN64-START reveals a hydrophobic tunnel 26 Å × 12 Å × 11 Å in size that extends almost the entire length of the protein and is apparently large enough to bind a single molecule of cholesterol. Because of this, Tsujishita and Hurley3 proposed that StAR functions in transferring cholesterol to the inner mitochondrial membrane, acting as a cholesterol shuttling protein (Fig. 2). Several mutations that result in lipoid CAH were mapped onto the MNL64-START structure. Three of these mutations (E169G, R182L, A218V) are quite close to each other in the tertiary structure, even though they are separated in sequence. In fact, these positions reside within the hydrophobic tunnel, and Tsujishita and Hurley3 predict that the lipoid CAH causing mutations would destabilize the tunnel, supporting its importance in the function of the START domain.

There are, however, a number of previous observations that are not easily reconciled by the proposal that StAR acts as a cholesterol shuttling protein, moving cholesterol to the inner mitochondrial membrane one molecule at a time. For example, as mentioned above, some results have suggested that StAR can be active in steroidogenesis even without being imported into mitochondria9,10,11. In addition, once StAR is imported into a mitochondrion, it is processed from the 37 kDa precursor to the 30 kDa mature form that is no longer active in cholesterol transfer. Therefore, it appears that StAR would not be able to act as a carrier protein on a continuous basis17 and because of this, one must consider whether enough cholesterol molecules could be delivered one at a time to P450scc, to account for the level of steroids synthesized, before the pool of StAR molecules becomes inactive. Moreover, mitochondrial proteins are required to be in an unfolded state before import can occur, and it is not clear how the refolding of StAR could occur in the intermembrane space, where it acts as a shuttling protein, because the refolding factors are thought to reside in the matrix.

Nevertheless, if StAR alone can indeed directly transfer cholesterol to the inner membrane, this may account for the inability of several investigators to identify StAR binding partners on the mitochondrial membrane. While many questions about StAR remain to be answered, Tsujishita and Hurley's3 accomplishment — determining the structure of a domain that is functional in cholesterol transfer — is a significant step toward solving the riddle of its mechanism of action.