MLN64 contains a domain with homology to the steroidogenic acute regulatory protein (StAR) that stimulates steroidogenesis

cDNAs and Plasmids.

A full-length MLN64 cDNA was prepared by reverse transcriptase PCR from human placental RNA and cloned into pCMV5. cDNAs encoding the wild-type MLN64 and deletion mutants were cloned into the pAT4 expression vector containing the F domain of the human estrogen receptor to epitope tag the proteins (7). A plasmid expressing human P450scc and its electron transport chain as a fusion protein (9) was generously provided by Walter L. Miller (University of California, San Francisco). The human StAR cDNA in pCMV5 has been previously described (6). Mutations in the human StAR cDNA were produced by site-directed mutagenesis as previously described (5).

Assessment of Steroidogenic Activity.

COS-1 cells were transfected with the plasmid expressing the cholesterol side-chain cleavage enzyme and the plasmids expressing MLN64, MLN64 mutants, human wild-type StAR or StAR mutants, or the empty vectors by using Lipofectamine as previously described (3, 4, 6). Cultures were incubated in the absence or presence of (22R)-22-hydroxycholesterol (5 μg/ml) for 36 h. Pregnenolone production was determined by radioimmunoassay using an antibody kindly provided by Charles Strott (National Institutes of Health) (6).

Experiments were performed on at least three separate occasions with three dishes in each treatment group. Pregnenolone production in the absence of exogenous hydroxysterol was taken to reflect the steroidogenic activity of the expressed proteins. The normalization of values to pregnenolone production in the presence of hydroxysterol, which reflects the total cholesterol side-chain cleavage activity, controls for variations in expression of this enzyme (4).

Western Blotting.

To verify expression of MLN64, the mutant MLN64 proteins, StAR, and StAR mutants, whole-cell extracts of transfected COS-1 cells were subjected to Western blot analysis using standard methods (5). Extracts of human placenta, BeWo choriocarcinoma cells, and fetal adrenal cortex were also examined. A monoclonal antibody (2BE2F4) raised against a peptide corresponding to 16 amino acid residues in the C terminus of MLN64 (amino acid residues 369–384) was used to detect MLN64 (7). A polyclonal antibody was also generated in rabbits against the same peptide. Monoclonal antibody F3A6 was used to detect the epitope-tagged fusion proteins (7). A polyclonal antibody generated against human recombinant StAR was used to detect StAR in COS-1 cell extracts (10).

Northern Blotting.

Membranes containing poly(A)⁺ RNA isolated from human adult and fetal tissues were purchased from CLONTECH and probed with an MLN64 cDNA as previously described (6). Total RNA was isolated from human granulosa-lutein and thecal cells and subjected to Northern blotting. A housekeeping gene, RLP 32 cDNA, was used to probe blots to establish loading of the lanes.

Identification of MLN64 and StAR Homologs.

The align program (11) was used to quantify the similarity between protein sequences. For the analysis reported here, 10,000 random permutations were used for the statistical analysis and the Dayhoff matrix was used wth a bias of 6 and a gap penalty of 8.

The Feng–Doolittle algorithm (12) was used to construct the phylogenetic tree of MLN64, StAR, and their homologs. The method of Fitch and Margoliash (13) was then used to obtain the branching order for the sequences. Branch lengths are calculated by a linear regression analysis of the best fit of the pairwise distances and the branching order. The lengths of the branches are proportional to the relative distance between the sequences.

Statistical Analysis.

Values presented are means ± SE of the indicated number of separate experiments. The paired t test or Student–Newman–Keuls test was used to determine significant differences among treatment groups, using P < 0.05 as the level of significance.

MLN64 Stimulates Steroidogenesis.

To test the steroidogenesis-enhancing activity of MLN64, we transfected monkey kidney COS-1 cells with the human cholesterol side-chain cleavage system and MLN64 and measured pregnenolone secretion. MLN64 caused an approximately 2-fold increase in pregnenolone secretion over COS-1 cells transfected with the cholesterol side-chain cleavage enzyme and empty plasmid vector (P < 0.01) (Fig. 1A). In contrast, human StAR increased pregnenolone production by nearly 7-fold. Thus, MLN64 demonstrated approximately 28% of the steroidogenic activity of StAR in this system.

MLN64 expressed in the COS-1 cells was detected as a predominant 50-kDa protein when a specific monoclonal antibody was used. Lower molecular mass proteins of about 42 and 33 kDa were also detected, probably representing proteolytic cleavage products (Fig. 1B a). MLN64 was not detected in the monkey kidney COS-1 cells transfected with empty vector, presumably because the monkey protein is not recognized by the monoclonal anti-MLN64 antibody. However, a polyclonal antibody detected MLN64 in the COS-1 cells transfected with empty vector, but at markedly lower levels than in cells transfected with the MLN64-expression vector (Fig. 1B b). To determine if the 50-kDa MLN64 and the smaller immunoreactive forms of the protein are present in tissues in vivo, we performed Western blot analysis on extracts of human placenta, BeWo choriocarcinoma cells, and fetal adrenal cortex. Fetal adrenal cortex contained 50-kDa and 33-kDa MLN64 species, whereas the placenta and BeWo extracts contained multiple lower molecular mass forms, including those of 42 and 33 kDa (Fig. 1B c). It should be noted that the anti-MLN64 antibodies do not crossreact with human StAR, nor do our anti-human StAR antibodies recognize MLN64 or the lower molecular mass MLN64 species.

The C Terminus of MLN64 Containing Sequences Homologous to StAR Is Required for Steroidogenic Activity.

To identify the domains of MLN64 that are necessary for steroidogenic activity, we examined mutant proteins in which either N-terminal or C-terminal sequences were deleted (Fig. 2). Deletions of the C terminus containing sequences homologous to StAR resulted in the complete loss of steroidogenic activity. In contrast, removal of N-terminal sequences, which contain the putative N-terminal microsomal targeting sequence and the putative transmembrane domains, increased steroidogenesis-enhancing activity to levels that are about two-thirds of the level observed with wild-type human StAR (ΔN2 relative steroidogenic activity: 4.6; wild-type StAR relative steroidogenic activity: 6.7). Each of the mutant proteins was detectable in the transfected COS-1 cells at its expected molecular mass (Fig. 3).

MLN64 Is Widely Expressed in Human Tissues.

To determine the pattern of expression of MLN64, we analyzed Northern blots containing poly(A)⁺ RNA from various human tissues (Fig. 4). The major 2.1-kb and minor 3.4-kb MLN64 mRNAs were detected in all adult tissues. Of particular note was the expression of MLN64 mRNA in placenta and brain, tissues that produce steroid hormones. Human fetal brain, lung, liver, and kidney contained the 2.1- and 3.4-kb mRNAs and some larger transcripts. Northern analysis also revealed the presence of the 2.1-kb MLN64 message in human granulosa and thecal cells (data not shown).

Homologs of MLN64 and StAR in Caenorhabditis elegans: Evolutionary Relationship of StAR and MLN64.

We previously reported that human MLN64 has a 478-residue putative homolog, F26F4.4, in C. elegans (7). To investigate further the origins of StAR, we did a series of blast searches of GenBank, which identified a 259-residue protein, F52F12.g (GenBank accession no. Z83228), that is homologous to human StAR based on an align analysis, which yielded a comparison score for the two proteins of 10.4 standard deviations higher than that for 10,000 comparisons of their randomized sequences. The probability of getting such a score by chance is 10⁻²³, strongly supporting the common ancestry of C. elegans F52F12.g and StAR. Interestingly, the comparison of F52F12.g with the StAR domain on human MLN64 and C. elegans F26F4.4 yields scores of 6.3 and 5 standard deviations, respectively, indicating that F52F12.g is closer to StAR than to MLN64 and that the StAR domains on these two proteins have diverged substantially. Fig. 5 presents a multiple alignment of mouse and human MLN64, StAR, and the two C. elegans proteins.

We further investigated the evolutionary relationship between the mammalian StAR proteins and their C. elegans homologs by constructing a phylogenetic tree using the Feng–Doolittle algorithm. As seen in Fig. 6, the two mammalian StAR proteins cluster on one branch, while the two mammalian MLN64 proteins cluster on another branch. The branch containing C. elegans F52F12.g is closer to the StAR proteins than to the MLN64 proteins, in agreement with the align scores.

The identification of the C. elegans 259-residue putative homolog of StAR and the 478-residue protein with a domain putatively homologous to StAR suggests that there was an ancient fusion of at least two genes to form MLN64. Considering that StAR and MLN64 arose over 600 million years ago, it is interesting that there are many positions with identical residues in either five or six StAR domains of the mammalian and invertebrate proteins (Fig. 5). As described below, the conserved motifs appear to be functionally important.

Mutations in Residues That Are Identical or Similar in StAR and MLN64 Result in the Loss of Steroidogenic Activity.

To determine if motifs conserved in StAR and MLN64 are important for functional activity, we categorized mutations that we previously described (3, 4, 5, 10) as well as nine newly created mutations according to whether they affected the steroidogenic activity of human StAR and if they were in residues that were identical, similar, or not conserved between StAR and MLN64. All but one of the mutations (H270Y) in human StAR residues that were identical or similar in MLN64 resulted in loss of steroidogenic activity (Table 1). The loss of activity ranged from complete to partial (100% to 19% inactivation). Although mutations in some residues that are conserved in StAR, MLN64, and the C. elegans proteins caused complete inactivation of StAR (e.g., R182L, ΔR272), mutations in other conserved residues (e.g., D246A and K248M) caused a lesser reduction in functional activity. In contrast, mutations of four different nonconserved residues had no significant impact on steroidogenesis-enhancing activity as did a four amino acid deletion of nonconserved residues in the C terminus. Western blot analyses of COS-1 cells transfected with constructs containing the newly created mutations that resulted in loss of steroidogenic activity demonstrated that the mutant proteins were expressed, with detection of the mature protein, indicating that the mutant StAR proteins can undergo mitochondrial import and processing (Fig. 7).