Regulation of Estradiol Synthesis by Aromatase Interacting Partner in Breast (AIPB)

ABSTRACT

Estradiol is essential for the development of female sex characteristics and fertility. Postmenopausal women and breast cancer patients have high levels of estradiol. Aromatase catalyzes estradiol synthesis; however, the factors regulating aromatase activity are unknown. We identified a new 22-kDa protein, aromatase interacting partner in breast (AIPB), from the endoplasmic reticulum of human breast tissue. AIPB expression is reduced in tumorigenic breast and further reduced in triple-negative tumors. Like that of aromatase, AIPB expression is induced by nonsteroidal estrogen. We found that AIPB and aromatase interact in nontumorigenic and tumorigenic breast tissues and cells. In tumorigenic cells, conditional AIPB overexpression decreased estradiol, and blocking AIPB availability with an AIPB-binding antibody increased estradiol. Estradiol synthesis is highly increased in AIPB knockdown cells, suggesting that the newly identified AIPB protein is important for aromatase activity and a key modulator of estradiol synthesis. Thus, a change in AIPB protein expression may represent an early event in tumorigenesis and be predictive of an increased risk of developing breast cancer.

INTRODUCTION

Aromatase (estrogen synthetase; EC 1.14.14.1) catalyzes the demethylation of androgens. It is most widely known for its role in reproduction and reproductive diseases and as a target for inhibitor therapy in estrogen-sensitive diseases, including endometriosis and leiomyoma (1, 2), as it catalyzes the conversion of testosterone to estradiol (3). Although aromatase is a microsomal protein present throughout the vertebrate phylum, the human gene is unique compared with the rest of the superfamily; it includes one untranslated 5′ exon 1 that modulates tissue-specific expression (4). Because aromatase is central to breast cancer development (5), a further understanding of aromatase control mechanisms may advance diagnostic and prognostic approaches.

Aromatase is expressed mainly in the ovaries of premenopausal women and placentas of pregnant women. In postmenopausal women, adipose tissue and skin cells are the major sources of estrogen production, but the aromatase activity in these tissues is significantly lower than in ovaries; thus, the level of circulating estrogen is much lower in postmenopausal women than in premenopausal and pregnant women. Since aromatase is responsible for the synthesis of estrogens, it is essentially a rate-limiting enzyme (6). Abnormal expression of aromatase in breast cancer cells and/or surrounding adipose stromal cells, especially in postmenopausal women, may have a significant influence on tumor maintenance and growth in breast cancer patients, given that estrogen is a major mitogenic stimulus to established breast cancer (7–10). Estrogen sources include ovaries, extraglandular sites, and breast tissue; however, the source responsible for the maintenance of estrogen concentrations in benign and breast cancer tissues remains unclear. It is most striking that the estradiol level in postmenopausal women is significantly higher than in the premenopausal state, and the mechanism of increased estradiol is still unexplained (11). In addition, postmenopausal women have higher concentrations of circulating testosterone in the blood than estradiol (11). Thus, it is essential to understand the mechanism of action of aromatase for enhanced estradiol synthesis during cancer progression and in postmenopausal women.

The most direct means of controlling breast cancer is to simply reduce estrogen by interfering with its production, via ovarian ablation in premenopausal women and use of aromatase inhibitors or inactivators in postmenopausal women. Although inactivators are a frequently used treatment for breast cancer in postmenopausal women, they have adverse effects due to depletion of estrogen, and some patients relapse (12), resulting in drug resistance and refractory disease. Identification of the mechanisms of resistance may provide predictive response markers and more effective treatment for hormone-dependent breast cancer. In spite of the markedly different circulating levels of estrogens in pre- and postmenopausal women, the concentration of estradiol in breast cancer tissues remains high. Although aromatase is known as the rate-limiting enzyme for estradiol synthesis, little is known about how it is modified posttranslationally and about the possible implications of these modifications for inactivator-resistant tumors that overexpress aromatase (13). Here, we describe a small 22-kDa protein, aromatase interacting partner in breast (AIPB), that was isolated from nontumorigenic breast tissue and is also expressed in tumorigenic breast tissue, though at a reduced level, with minimal expression in Er (estrogen receptor)-negative or triple-negative tumors. AIPB is stimulated by estrogen similarly to aromatase; it interacts strongly with aromatase in the endoplasmic reticulum (ER) in both tumorigenic and nontumorigenic breast tissue. When we blocked AIPB availability in breast cells or tissues, estradiol synthesis was increased, and similarly, conditional overexpression of AIPB decreased estradiol synthesis. In the absence of AIPB, estradiol synthesis increased severalfold in tumorigenic cells, suggesting that AIPB is a newly identified partner for aromatase and a key modulator for the control of estradiol synthesis in the human breast during tumorigenesis.

RESULTS

Identification of a new ER-associated protein in the breast.

The metabolic conversion of testosterone to estradiol requires the cofactor NADPH and aromatase, but the need for any interacting partner to regulate the catalysis is unknown. Aromatase resides in the ER, where it may interact with a partner to regulate estradiol synthesis (3, 14). We first determined metabolic activity of ER fractions isolated from tumorigenic and nontumorigenic human breast tissue. Analysis of the metabolic activity of ER fractions from the same patient showed high estradiol production in tumorigenic breast tissue compared to nontumorigenic tissue (921.2 pg/ml versus 173.5 pg/ml; P = 0.019) (Fig. 1A and B). Similar results were observed with an increase in estradiol synthesis in the ER fraction of tumorigenic T-47D or nontumorigenic MCF-12A cells (389 pg/ml versus 165.9 pg/ml; P = 0.023) (Fig. 1A and B). Western blotting with a calnexin antibody showed a similar level of expression in each reaction, indicating the presence of the same amount of total protein in each of the metabolic reactions (Fig. 1A, bottom). Gas chromatography-mass spectrometry (GC-MS) analysis confirmed estradiol synthesis (see Fig. S1A and B in the supplemental material), substrate testosterone (Fig. S1C and D), and minimal production of the by-product estrone (data not shown). These data suggest that nontumorigenic tissue expresses a specific factor that reduces aromatase activity, resulting in lower estradiol production.

FIG 1.

Estradiol synthesis is regulated by an aromatase interacting partner in breast. (A) (Top) Metabolic conversion of testosterone to estradiol with the ER fraction and its analysis through TLC. (Bottom) Western blot with calnexin antibody of the endoplasmic fraction applied in the reaction. (B) Measurement of estradiol synthesis from the ER isolated from tumorigenic and nontumorigenic human breast tissue. The estradiol measurement from a tumorigenic and nontumorigenic breast tissue as well as from the nontumorigenic (MCF-12A) and tumorigenic (T-47D) cells. Data are means plus standard errors of the means (SEM) from three independent experiments performed at three different times. (C) Schematic presentation of the SDS-PAGE fraction, showing bands excised for mass-spectrometric analysis. (D to G) LC-MS/MS analysis of the peptide from the section III region of nontumorigenic (D) and tumorigenic (E to G) breasts. The red arrow shows the retention time of 21.15 min of the newly identified peptide. (H) Identification of a peptide sequence from breast tissues. (I to K) PCR amplification of the cDNA by 5′ (I) and 3′ (J) RACE followed by generation of full-length cDNA (K). (L) The newly cloned protein sequence. Red letters or arrows show the peptide found from the mass spectrometric analysis that was present in the open reading frame. (M) cDNA sequence analysis. The red box shows the mass spectrometry-identified protein sequence (see panel L) which was included during full-length cloning of the AIPB cDNA, with extension of the newly generated cDNA through 5′ and 3′ RACE. The open reading frame (ORF) contains 621 bp and also an untranslated region of 306 bp.

Tissue-specific aromatase expression depends on different factors (15). Because aromatase catalyzes estradiol synthesis at the ER of breast tissue, we hypothesized that any regulating protein may also be localized in the same organelle. After isolating and purifying the ER from nontumorigenic and tumorigenic (Er⁺/Pr⁺/Her²⁻, Er⁻/Pr⁻/Her²⁺, and triple-negative) breast tissues, protein expression was first divided into three sections (I, II, and III) and was then excised in every 10-kDa fraction from SDS-PAGE (Fig. 1C) for accurate detection of the presence of new protein (16, 17). All the proteins identified by liquid chromatography-tandem mass spectrometry (LC-MS/MS) (Fig. 1D to G) were already known in ER breast tissue, except one specific peptide, LEVVVDQPMER (Fig. 1H and Table S1). The identified peptide (Fig. 1H, red arrow) is homologous to a mitochondrial resident cholesterol trafficker (CT) present mainly in adrenals and gonads but completely absent in the breast (18). The peptide was observed less than 50% of the time in double- and triple-negative tumors.

Identification of aromatase interacting partner.

To characterize the entire sequence of the identified peptides, we proceeded to clone the complete cDNA by 5′ and 3′ rapid amplification of cDNA ends (RACE) from RNA derived from nontumorigenic human breast tissue. The 3′ sequencing resulted in a completely new cDNA sequence (Fig. 1I) with a stop codon at 361 bp and a poly(A) tail. Next, we performed 5′ RACE, generating an additional 450 bp cDNA (Fig. 1J). The identified 5′ and 3′ RACE-amplified sequence was cloned (Fig. 1K), resulting in a 621-bp open reading frame, which showed a 207-amino-acid protein not present in the NCBI database (Fig. 1L). The mass spectrometry-identified sequence is located at positions 133 to 143 (Fig. 1L, red). The cDNA has 306 bp of the untranslated region (UTR) at the C terminus after the open reading frame (shown schematically in Fig. 1M) (GenBank no. MT920320) (19). The newly identified protein from human breast was named aromatase interacting partner in breast (AIPB).

We next performed sequence analysis. The first 42 amino acids of AIPB have identity with the N-terminal amino acids of CT. Further sequence analysis showed that the total 57 amino acids of AIPB have identity with the 285 amino acids of CT, which is about 20% (Fig. 2A). However, CT is expressed mainly in the adrenals and gonads (ovaries for women) and minimally in the brain (3, 20), but it is absent in breast tissue (18) (Fig. 2B). To further confirm the specificity of cDNA, we amplified AIPB with the cDNA prepared from MCF-12A and T-47D RNA but not from human ovary RNA or from genomic DNA (Fig. 2C, left). As a control, the flap structure-specific endonuclease gene (FEN1) ubiquitously expressed in many tissues was expressed in all the synthesized cDNA samples, confirming the accuracy of cDNA synthesis (Fig. 2C, right). To confirm the specificity of AIPB expression, we performed Western blotting of the nontumorigenic MCF-12A cells and monkey kidney COS-1 cells. In MCF-12A cells, a 22-kDa protein was found that matched the molecular weight of AIPB; however, there was no AIPB expression in COS-1 cells (Fig. 2D), but aromatase was found in MCF-12A and calnexin in both cell types (Fig. 2D, middle and bottom). For further confirmation, we did Western blotting with an antibody specific for the C terminus only of CT, which resulted in a band of approximately 30 kDa from MA-10 and NCI-H295 cells, but no expression was found with breast nontumorigenic MCF-12A or monkey kidney COS-1 cells (Fig. 2E, top). Analysis of calnexin was used as a loading control (Fig. 2E, bottom).

FIG 2.

Identification and homology comparison of AIPB protein. (A) Homology comparison between the newly generated AIPB protein with cholesterol trafficker (CT) protein. Dashes represent gaps, and dots represent semiconserved substitutions (similar residues). The dots show that we were unable to find a match in that given region or the match did not pass the threshold. The red solid line from 196 to 208 shows the peptide identified by mass-spectrometric analysis and present in AIPB. (B) Confirmation of the identification of the specificity of cDNA AIPB through RT-PCR prepared from breast tissue with CT primer and AIPB primer. Additional PCR amplification was performed from the total RNA of the indicated resource. (C) (Left) Confirmation of the identification of the specificity of cDNA AIPB through RT-PCR prepared from nontumorigenic MCF-12A and tumorigenic T-47D cells and a parallel comparison with cDNA prepared from human ovary. The results were compared with the parent AIPB cDNA and in the absence of reverse transcriptase. (Right) RT-PCR of the housekeeping gene FEN1, which is also a nuclear gene with no introns interrupting the protein coding sequence. Both MCF-12A and T-47D cells amplified the FEN1 product along with genomic DNA. (D) (Top) Western blot analysis of expression pattern of AIPB in breast cells and comparison with monkey kidney (COS-1) cells with a CT antibody. (Middle and bottom) Expression of aromatase (middle) and calnexin (bottom) antibodies independently. (E) Western blot with the C-terminus-specific CT antibody applied to the same amount of total cell lysate prepared from the indicated cells. (Bottom) Western blot with the same lysate with calnexin antibody, showing the presence of the same amount of total protein applied in each reaction.

AIPB is essential for estradiol synthesis.

Aromatase catalyzes testosterone conversion to estradiol in breast tissue (Fig. 1A). To understand the physiological relevance of AIPB in estradiol conversion, we measured estradiol synthesis after knocking down its expression by small interfering RNA (siRNA) in MCF-12A (Fig. 3A) and T-47D (Fig. 3B) cells. AIPB-specific siRNAs reduced expression up to 86% (Fig. S2A and B) without affecting aromatase expression (Fig. 3A and B, middle), confirming that AIPB expression is independent of aromatase. Next, we determined testosterone to estradiol conversion following AIPB siRNA knockdown in MCF-12A and T-47D cells. As shown in Fig. 3C, there was minimal estradiol conversion in MCF-12A cells in the absence of siRNA or nonspecific siRNA; however, estradiol levels increased following AIPB knockdown (173.6 versus 54.1 pg/ml). Estradiol level was also significantly increased following AIPB knockdown in T-47D cells (110.8 pg/ml versus 443 pg/ml) (Fig. 3C). Calnexin expression was unaltered following AIPB knockdown, suggesting that AIPB siRNA did not affect the expression of other ER proteins (Fig. 3C, bottom). Using three siRNAs specific for CT reduced its expression in mouse Leydig (MA-10) cells (Fig. 3D and Fig. S2C), but the same siRNAs remained ineffective in MCF-12A cells (Fig. 3E). The expression of the ER resident protein calnexin remained unchanged, suggesting that CT is not present in MCF-12A cells.

FIG 3.

Identification of siRNA for reducing AIPB expression and estradiol synthesis. (A and B) (Top) MCF-12A (A) and T-47D (B) cells were incubated with the indicated siRNA oligonucleotides, and AIPB expression was determined by Western blotting with a CT antibody. siRNA-transfected cells were analyzed with aromatase (middle) and calnexin (bottom) antibodies independently. (C) Measurement of estradiol synthesis following knockdown of AIPB cDNA by siRNA in MCF-12A and T-47D cells. (Top) Quantitative analysis of the amount of estradiol synthesized. (Bottom) Western blot of the ER proteins used for enzymatic analysis in the top panel by calnexin antibody. (D and E) Effect of CT knockdown by siRNA oligonucleotides in MA-10 and MCF-12A cells. Three different siRNA oligonucleotides (30 pmol) were incubated with MA-10 (mouse Leydig) cells (D) and MCF-12A cells (E), and the expression was determined by Western blotting with a CT antibody. (Bottom) Western blots of the siRNA-treated cells using a calnexin antibody. (F) (Top) Stable AIPB expression in tumorigenic MCF-7 cells and MCF-7 cells transfected with a GFP-expressing plasmid and no transfection of HEK-293 cells. AIPB expression was detected using a CT antibody. (Second from top) Western blot of the same lysates with a GFP antibody, showing similar levels of GFP expression in both cells. (Third from top) Western blotting of the same lysates with an aromatase antibody, showing similar aromatase expression. (Bottom) Western blot of the same lysates with calnexin antibody. (G) (Top) Measurement of estradiol synthesis by the AIPB stable cells with and without induction of doxycycline and GFP transfection. The activity was compared with that in the MCF-7 cells with and without GFP transfection. (Bottom) Western blotting of the same cell lysates with (top to bottom) GFP, CT, and aromatase antibodies independently, showing the presence of same amount of total protein applied in each experiment. (H) Localization of aromatase (Ar) by immuno-EM staining with an aromatase antibody (red arrow). (I) Colocalization of AIPB (15 nm; cyan arrows) and aromatase (55 nm; red arrows). (J) Localization of AIPB by immuno-EM staining with a CT antibody (cyan arrows). (K) Similar localization by ER-resident GRP78 antibody. (Left) Complete tissue section; (right) magnification of the boxed section on the left. Bars, 1.0 μm (H and J) and 0.5 μm (I and K). (L) Semiquantitative analysis of Ar and AIPB colocalization from panel I. (M) Localization of AIPB in the organelle fractions of tumorigenic T-47D cells. (Top) Purified ER, MAM, and mitochondrial fractions from tumorigenic T-47D cells were analyzed by Western blotting with a CT antibody. (Bottom) Western blotting of the same lysates in the top detected by calnexin and VDAC2 antibodies independently. (N) Summary of the results showing the presence of AIPB along with aromatase-reduced estradiol level (downward black arrow), whereas in the absence of AIPB, estradiol level is elevated severalfold (red arrows). Data in panels B, C (top), G, and H are means plus SEM from three independent experiments performed at three different times.

To confirm that AIPB is responsible for reducing estradiol synthesis, we cloned it into a doxycycline inducible vector and selected stable clones expressed in MCF-7 cells using hygromycin. On incubation of doxycycline, the expression of AIPB was significantly higher in stable clones than in wild-type MCF-7 cells (Fig. 3F, top). As a control, we also applied the same amount of protein from HEK-293 cells, which do not express aromatase and AIPB. We transfected the AIPB-stable cells and MCF-7 cells independently with a humanized Renilla green fluorescent protein (hrGFP) expression vector. The fusion GFP results in a 28-kDa protein. Western blotting with the GFP antibody showed a similar level of GFP (28-kDa protein) in both the cells (Fig. 3F, second panel from top). As expected, aromatase (Fig. 3F, third panel from top) and calnexin (Fig. 3F, bottom) expression remained unchanged. Estradiol levels of the AIPB-stable cells before induction with doxycycline were 463 pg/ml, which decreased to 223.8 pg/ml following induction (Fig. 3G, top). Estradiol synthesis in the MCF-7 cells was about 379 pg/ml with and without GFP transfection (Fig. 3G, top). The activities of the vector and the heat-inactivated MCF-7 cells were minimal. Western blotting with a GFP antibody showed the presence of a similar amount of total protein applied in GFP-transfected cells (Fig. 3G, second from top). The expression of AIPB was similar in all the cells and increased with addition of doxycycline (Fig. 3G, third from top). The endogenous aromatase expression was almost identical in all the reactions (Fig. 3G, bottom). These results suggest that increasing AIPB expression reduces estradiol synthesis.

Following AIPB overexpression in MCF-7 cells, estradiol synthesis was reduced (Fig. 3G), suggesting that it interacts directly with aromatase. Thus, these two proteins may be colocalized. Analysis of nontumorigenic breast tissue by immuno-electron microscopy (immuno-EM) showed that AIPB was mostly localized in the ER and minimally localized near the outside of mitochondria, whereas aromatase was localized in the ER (Fig. 3H and J; the right panels are magnifications of the boxed regions in the left panels). To confirm whether these two proteins are closely localized in the ER and to determine their relative expression in the same organelle, we performed colocalization analysis probing with AIPB (15 μm AIPB) (Fig. 3J, cyan arrows) and aromatase (55 μm aromatase) (Fig. 3H, red arrow) antibodies together. Immuno-EM showed that aromatase (Fig. 3I, red arrows) and AIPB (Fig. 3I, cyan arrows) were present at the ER. The ER resident GRP78 was localized in the ER (Fig. 3K; the magnified ER section is on the right). A quantitative analysis of the 40 best images showed 13 ± 0.9 aromatase overlapping 11 ± 1.93 AIPB per 81-μm² field of view in the ER, suggesting an overlap of 85% between the two proteins (Fig. 3L). Fractionation of tumorigenic T-47D cells followed by Western blotting with CT antibody showed 22-kDa AIPB localized predominantly at the ER fraction (Fig. 3M). A minor, less intense band of about 12 kDa at the mitochondrion-associated ER membrane (MAM) is possibly a breakdown of AIPB. Western blotting with an ER resident, calnexin, and a mitochondrial resident, VDAC2 (21), confirmed the accuracy of fractionation. In summary, suppression of AIPB expression increased estradiol synthesis in tumorigenic cells without affecting aromatase expression (Fig. 3N), possibly because of their close localization in the ER, suggesting that AIPB may regulate estradiol levels in the breast.

AIPB expression is dependent on Er and Pr expression.

Estradiol synthesis requires androgenic substrate from endocrine, paracrine, or autocrine sources (22). Given that AIPB knockdown increased estradiol synthesis, we hypothesized that it may have a role during tumorigenesis in the breast. Therefore, we determined AIPB expression in nontumorigenic and different tumorigenic human breast tissues by Western blotting. AIPB is expressed in the nontumorigenic breast tissue as a 22-kDa protein; its expression was reduced in Er⁻/Pr⁻/Her²⁻ tumors and even more so in other tumors (Fig. 4A). Quantitative analysis showed that AIPB expression was reduced by 36% in the Er⁻/Pr⁻/Her²⁻ tumors and up to 86% in the others (Fig. 4B), suggesting that AIPB expression is dependent on the presence of Er and Pr. The expression of the bottom 17-kDa band was unchanged in all tissues, as seen on longer exposure (Fig. 4A, right), suggesting that it is possibly a breakdown product of the 22-kDa protein. Western blotting of the same lysates with calnexin antibody shows similar expression in all lanes, confirming the presence of a similar amount of total protein (Fig. 4A, bottom left).

FIG 4.

Regulation of AIPB expression in human nontumorigenic and tumorigenic breast tissue. (A) Expression pattern of AIPB in nontumorigenic and different tumorigenic breast tissues. (Top) Western blotting of the indicated tissues with a CT antibody. The expression pattern resulted in 22-kDa and 17-kDa bands. The expression of the 17-kDa protein remained unchanged across all tissues, but expression of the 22-kDa band changed, as shown by longer exposure (right). (Bottom) Expression of calnexin in the various tissues. (B) Quantitative analysis of the expression pattern of AIPB from panel A. (C) (Top) Quantitative analysis of the change in expression of AIPB (22 kDa) with increasing concentration of cAMP. (Bottom) Western blot with calnexin antibody showing the same amount of calnexin. (D) Effect of cycloheximide (CHX) on AIPB expression in the presence and absence of cAMP. (E) COS-1 cells stably expressing AIPB were incubated with 0.5 mM cAMP for the indicated time, and AIPB expression was detected by Western blotting with CT antibody. (Bottom) Western blotting with calnexin antibody of the same amount of lysate at each time point of the top panel. (F) Effect of the indicated concentrations of α-zeranol on the expression of aromatase in MCF-12A and MCF-7 cells. (G) Effect of α-zeranol on AIPB expression in MCF-12A and MCF-7 cells by Western blotting. (H to J) Western blotting of AIPB (H), aromatase (I), and calnexin (J) expression levels with increasing concentrations of cell lysate in MCF-12A and MCF-7 cells. Data in panels B to D and F to J are means and SEM from three independent experiments performed three times.

AIPB and aromatase possess similar biochemical properties.

In ovary and breast tissue, the estrogen activator PGE2 stimulates estradiol synthesis by increasing aromatase expression (1, 23). However, AIPB is absent in the ovary (Fig. 2C). Most steroidogenic proteins are activated acutely or transcriptionally, with the exception of aromatase. To understand the mechanism of AIPB activation in breast cells, we stimulated MCF-12A cells with various cAMP concentrations for 24 h and then determined expression by Western blotting. Although AIPB expression was unchanged with 0.5 mM cyclic AMP (cAMP), it decreased with higher cAMP concentration (Fig. 4C). The decrease in AIPB expression may be due to cAMP toxicity at higher concentrations; however, calnexin expression was similar, confirming the presence of a similar amount of total protein (Fig. 4C, bottom). We next determined transcriptional activation by incubating MCF-12A cells with cycloheximide in the presence and absence of 0.5 mM cAMP, but AIPB expression was unchanged under both conditions (Fig. 4D). Next, we stably selected AIPB expression in nonsteroidogenic COS-1 cells in the presence of G-418. The expression of AIPB was unchanged with incubation of 0.5 mM cAMP from 4 h to 48 h (Fig. 4E), suggesting that AIPB is not stimulated by cAMP. Taken together, these results suggest that AIPB is similar to aromatase in that it is not a cAMP-regulated protein in breast cells.

Aromatase is not stimulated by cAMP but requires an estrogenic stimulator for activation in breast cells (24–26). To understand if AIPB also requires an estrogenic stimulator, we incubated nontumorigenic MCF-12A and tumorigenic MCF-7 cells with a 100-fold difference in concentration of α-zeranol, a nonsteroidal anabolic growth promoter with estrogenic activity that activates aromatase expression (27, 28). Aromatase expression in nontumorigenic MCF-12A cells increased 2.3-fold in 50 nM α-zeranol compared to dimethyl sulfoxide (DMSO)-treated control cells (Fig. 4F), while its expression increased a further (3.6-fold) in tumorigenic MCF-7 cells (Fig. 4F). Similarly, the expression of AIPB increased 2.5-fold in nontumorigenic MCF-12A cells in the presence of 100 nM α-zeranol compared to untreated cells (Fig. 4G). In contrast, AIPB expression did not change significantly in tumorigenic MCF-7 cells treated with α-zeranol under identical conditions (Fig. 4G). These data suggest that AIPB and aromatase are regulated in a similar manner in nontumorigenic breast cells; however, the Er agonist α-zeranol is significantly ineffective in changing AIPB expression in tumorigenic cells.

We next hypothesized that AIPB and aromatase expression levels differ between MCF-12A and MCF-7 cells. Western blotting with increasing total protein concentrations showed that AIPB expression in MCF-12A cells increased with higher protein concentration; however, the increase was not pronounced in MCF-7 cells (Fig. 4H). In contrast, aromatase expression was significantly higher in MCF-7 cells than in MCF-12A cells (Fig. 4I). Quantitative analysis showed that AIPB expression is 2.3 times higher in MCF-12A cells than in MCF-7 cells with 2 mg/ml total protein, whereas aromatase expression in MCF-7 cells is 2.5-fold higher than in MCF-12A cells at the same concentration (Fig. 4H and I). The expression of calnexin in both cells under identical conditions increased linearly (Fig. 4J), confirming the accuracy of total protein concentration. Therefore, AIPB and aromatase expression are not proportionately increased in tumorigenic cells.

To understand the impact of a minimal increase in AIPB and aromatase expression in nontumorigenic MCF-12A cells compared to MCF-7 tumorigenic cells treated with α-zeranol for 24 h (Fig. 4F and G), we measured estradiol synthesis. Estradiol increased by 4-fold following induction of the MCF-7 cells (from 203 pg/ml to 855 pg/ml) (Fig. 5A), but a similar incubation of nontumorigenic MCF-12A cells resulted in a far smaller increase (135 pg/ml to 280 pg/ml) (Fig. 5A). To confirm further that the increase in activity is due to increased expression of aromatase, we measured aromatase activity with the ER isolated from the nontumorigenic (MCF-12A) and tumorigenic (MCF-7) cells by enzyme-linked immunosorbent assay (ELISA). The result shows that between 42 and 48 h, aromatase activity in MCF-7 cells reached almost 4-fold-higher levels than in MCF-12A cells (Fig. 5B). To understand maximum aromatase activity, we studied MCF-7 cell growth with 10, 25, 50, and 100 nM α-zeranol for up to 96 h. There was a minimal effect with 10 nM α-zeranol, and 100 nM was possibly toxic. As expected, MCF-7 cell growth (viability) was unchanged with 25 nM and 50 nM α-zeranol for almost 42 h and then started decreasing gradually with 50 nM more than with 25 nM. The growth was significantly reduced after 72 h (Fig. 5C), with a larger effect with 50 nM and less with 25 nM, suggesting that aromatase expression is directly associated with the growth of MCF-7 cells. The growth of AIPB-MCF-7 stable cells with α-zeranol or doxycycline did not change, possibly due to slow growth of AIPB stable cells or because synthesized estradiol level (Fig. 3G) is sufficient for the maintenance of cell growth (Fig. 5D and E). Since the expression of AIPB was very low in Er⁻/Pr⁻/Her²⁻ compared to Er⁺/Pr⁺/Her²⁻ breast tumors (Fig. 2A), we determined aromatase activity in MCF-7 (Er⁻/Pr⁻/Her²⁻) and MDA-MB-231 (Er⁻/Pr⁻/Her²⁻) cells and compared with the MCF-12A cells treated with 50 nM α-zeranol for 48 h. The activity of MCF-7 cells increased from 2.2 ng/ml to 4.89 ng/ml after α-zeranol induction (Fig. 5F). Similarly, aromatase activity in MDA-MB-231 cells increased from 5.89 ng/ml to 7.3 ng/ml after α-zeranol induction (Fig. 5F). As expected, the expression of AIPB in MCF-12A cells was significantly increased; it was minimally increased in MDA-MB-231 cells and moderately increased in MCF-7 cells after induction (Fig. 5G). In summary, these data suggest that the AIPB expression did not increase significantly compared to that of aromatase in tumorigenic cells; the higher aromatase levels favored greater estradiol synthesis.

FIG 5.

Effect of the estrogen stimulator α-zeranol on AIPB expression and estradiol synthesis. (A) Measurement of estradiol synthesis by RIA following a 24-h incubation of MCF-12A (nontumorigenic) and MCF-7 (tumorigenic) cells with 50 nM α-zeranol. (B) Measurement of aromatase activity by ELISA with the purified ER fraction after incubation of the nontumorigenic (MCF-12A) and tumorigenic (MCF-7) cells with 50 nM α-zeranol. (C) Measurement of the growth of MCF-7 cells in the presence of the indicated concentrations of α-zeranol from 12 to 96 h. (D and E) Growth analysis of MCF-7 cells, which are stable in the presence of 50 mM α-zeranol (D) and 100 μg/ml doxycycline (E), measured at the indicated times. (F) Measurement of aromatase activity by ELISA with the purified ER fraction after incubation of nontumorigenic (MCF-12A) and tumorigenic (MCF-7 and MDA-MB-231) cells with 50 nM α-zeranol for 48 h. (G) Measurement of AIPB expression in the absence and presence of α-zeranol. Data in all panels are means and SEM from three independent experiments performed three times.

AIPB and aromatase interact in breast.

Because of the similar locations of AIPB and aromatase at the ER of breast tissue (Fig. 3I), we hypothesized that they may interact directly. Coimmunoprecipitation (co-IP) analysis of tumorigenic and nontumorigenic breast tissues stained with aromatase (Fig. 6A and C) and CT (Fig. 6B and D) antibodies independently pulled out AIPB protein using both CT and aromatase antibodies but not the COX IV or VDAC2 antibodies or rat IgG, suggesting that AIPB and aromatase interact with each other in human breast tissue. Western blotting of part of the tissue lysate from the immunoprecipitation reaction with calnexin and aromatase or CT antibodies confirmed that an identical amount of protein was applied in each reaction (Fig. 6A to D, middle and bottom). An identical co-IP experiment with the MCF-12A (Fig. 6E and F) and T-47D (Fig. 6G and H) cells with aromatase (Fig. 6E and G) and CT (Fig. 6F and H) antibodies independently pulled out AIPB and aromatase but not IgG, COX IV, and VDAC2 antibodies, validating the specificity of the AIPB-aromatase interaction in breast tissue. Western blotting of the cell lysates applied in each reaction with calnexin and aromatase or CT antibodies showed similar expression (Fig. 6E to H, middle and bottom).

FIG 6.

AIPB and aromatase interaction. (A to H) Co-IP with the indicated antibodies followed by Western blotting with CT and aromatase antibodies independently in tumorigenic tissue (A and B), nontumorigenic breast tissue (C and D), MCF-12A cells (E and F), and T-47D cells (G and H). (Bottom) Western blotting with CT or aromatase and calnexin antibodies independent of the lysates prior to immunoprecipitation showing the presence of the same amount of lysate applied in each reaction. (I) (Left) Titration with increasing concentrations of cross-linker in MCF-12A cells through in vivo chemical cross-linking followed by identification by Western blotting with an aromatase antibody. (Right) Quantitative analysis of the chemical cross-linking in the left panel. (J to M) In vivo chemical cross-linking with 2 mM BS3 in MCF-12A (J and L) and T-47D (K and M) cells followed by Western blotting with CT (J and K) and aromatase (L and M) antibodies independently. (N) (Top) Measurement of the estradiol synthesis by RIA with the MCF-7 cell lysis after incubation of aromatase, CT antibodies, and IgG independently. (Bottom) Western blot of the cell lysates from the top panel with calnexin antibody, showing the presence of the same amount of protein applied in each reaction. Data in panels I (right) and N are means and SEM from three independent experiments performed three times.

AIPB and aromatase are closely associated.

The 3-dimensional crystal structure of aromatase shows hydrophobic and polar residues with 12 major α-helices and 10 β-strands. Extensive structure-function studies, through a combination of site-directed mutagenesis of the enzyme and inhibitor binding characterization by several scientists, revealed that the functions of the different inhibitors are not the same (29). Thus, AIPB and aromatase may interact to catalyze the conversion of testosterone to estradiol. To confirm the interaction between AIPB and aromatase, we performed in vivo chemical cross-linking, incubation with various concentrations of the homobifunctional cross linker BS3, and visualization by Western blotting with an aromatase antibody. The interaction started immediately (0.1 mM), resulting in a 79-kDa band (Fig. 6I, left), with maximal interaction at 2 mM (Fig. 6I, right) and decreased with higher concentrations, possibly due to cross-linker toxicity. To confirm the specificity of cross-linking, we performed co-IP analysis with ER fractions obtained from MCF-12A and T-47D cells cross-linked with 2 mM BS3 followed by staining with CT and aromatase antibodies independently. Western blotting again showed a 79-kDa cross-linked reaction product containing 22-kDa AIPB (Fig. 6J and 57-kDa K) and aromatase (Fig. 6L and M) but not IgG, COX IV, and VDAC2 antibodies, confirming a direct interaction between AIPB and aromatase proteins.

We next hypothesized that blocking the availability of AIPB with CT antibody would result in more aromatase availability and thus increase estradiol synthesis. We indeed observed that incubation of MCF-7 cell lysates with CT antibodies increased estradiol activity from 174 pg/ml to 439 pg/ml (Fig. 6N), but similar incubations with rat IgG had no effect. Additional confirmatory experiments under identical conditions, including incubation with aromatase antibodies, reduced estradiol synthesis minimally (174 pg/ml to 141 pg/ml), further corroborating our hypothesis that AIPB interacts in a mechanistic fashion with aromatase to reduce estradiol synthesis (Fig. 6N).

Specificity of AIPB-aromatase interaction.

The active site of aromatase is buried deep within the roughly spherical molecule near its geometric center (14), suggesting that specific regions of a protein are required for interaction. To understand the specific amino acids of AIPB required for interaction with aromatase, we performed deletional mutagenesis of AIPB. As a minimum of 12 amino acids are essential for targeting to the organelle (30), we performed internal deletional mutagenesis by deleting five amino acids at a time starting at residue 13. We generated Δ13–17, Δ13–22, Δ13–27, and Δ13–32 constructs, in which approximately 10% of the AIPB open reading frame had been removed (Fig. 7A). Next, we overexpressed each of the constructs along with full-length aromatase cDNA in nonsteroidogenic HEK-293 cells (Fig. 7B). Co-IP with IgG, COX IV, VDAC2, aromatase, and CT antibodies followed by Western blotting with an aromatase antibody showed that the interaction decreased following deletion of amino acids 13 to 17 (Fig. 7C). Further deletion of amino acids 13 to 22 or 13 to 27 disrupted the interaction with aromatase (Fig. 7D and E). The interaction was completely disrupted with the deletion of amino acids 13 to 32, suggesting that the internal sequence spanning amino acids 13 to 32 is essential for minimal interaction with aromatase (Fig. 7F). To confirm the accuracy of the transfection, we performed Western blotting of the reactions employed in the transfection experiments with a calnexin antibody, which showed similar levels of expression (Fig. 7C to G, bottom). The different AIPB deletion mutants also showed very similar levels of expression (Fig. 7G).

FIG 7.

Protein-protein interaction by co-IP. (A) Schematic presentation of the different internal deletional mutants. (B) AIPB cotransfected with aromatase in HEK cells. (C to F) Cotransfection of wild-type AIPB, internal deletion mutants of AIPB from the N terminus (Δ13–17 [C], Δ13–22 [D], Δ13–27 [E], and Δ13–32 [F]), and full-length aromatase cDNA in HEK-293 cells, followed by co-IP with the indicated antibodies and Western blotting with an aromatase antibody. (Bottom) Western blots of the cells in the top panel probed with a calnexin antibody. (G) (Top) Western blot of the full-length AIPB and internal deletional mutants transfected in panels C to F. (Bottom) Western blots of the cells in the top panel probed with a calnexin antibody, showing similar amounts of lysate present in all lanes.

DISCUSSION

Estradiol synthesis is dependent on the catalysis of testosterone by aromatase in the presence of the cofactor NADPH. As the concentrations of NADPH do not change in breast tumorigenesis, the mechanism responsible for increased estradiol synthesis in breast cancer has remained a mystery. The crystal structure of aromatase shows hydrophobic and polar residues with 12 major α-helices and 10 β-strands (14). Structure-function studies, using site-directed mutagenesis of the enzyme and inhibitor binding assays, revealed that the functions of the different inhibitors are not the same (29). Therefore, AIPB is likely an interacting partner of aromatase for estradiol synthesis. In tumorigenic cells, aromatase expression increased while AIPB expression was essentially unchanged, resulting in increased estradiol synthesis.

In this report, we describe a mechanism for regulation of aromatase activity by AIPB based on the following results. First, AIPB knockdown by siRNA resulted in increased aromatase activity (as measured by estradiol synthesis) (Fig. 3A and C). Second, upon stimulation with an estrogen stimulator, estradiol synthesis increased for a limited time, as shown by tumorigenic MCF-7 cells having higher aromatase expression than AIPB (Fig. 4F and G). Therefore, there is only a limited increase in estradiol synthesis under identical conditions in the nontumorigenic MCF-12A cells (Fig. 5A). Third, in a regulated overexpression of AIPB induced with doxycycline, aromatase activity was reduced (as measured by estradiol synthesis) (Fig. 3F and G). Fourth, blocking AIPB availability with CT antibody decreased the interaction with aromatase, resulting in an increase in estradiol synthesis (Fig. 6N). Fifth, aromatase expression was higher in MCF-7 cells than MCF-12A cells, which increased with an estrogen stimulator, resulting in a significant increase in estradiol synthesis (Fig. 4F and 5A). Sixth, aromatase and AIPB are both colocalized at the ER, as determined by immuno-electron microscopy (Fig. 3I, right). Seventh, Co-IP analysis showed stronger interactions between the two proteins in ER fractions isolated from nontumorigenic breast tissue than tumorigenic tissues (Fig. 6A to D). Similar results were observed in a cellular system (Fig. 6E to H). Eighth, the reduced level of interaction in tumorigenic cells supports the reduction in AIPB expression in these cells relative to aromatase (Fig. 6J to M). Aromatase is an ER-bound protein, and its folding may require lipid vesicles and anchoring with the membrane (31). Ninth, the specificity is likely due to an association of amino acids 13 to 27 of AIPB with aromatase (Fig. 7C to F). Therefore, a decrease in AIPB expression may alter aromatase-AIPB assembly in the membrane, thereby reducing or eliminating this key interaction.

In summary, we identified AIPB as a novel protein that works with aromatase to regulate estradiol synthesis. In the absence of AIPB, as in tumorigenic tissues, aromatase is overly available, resulting in high levels of estradiol synthesis. AIPB interacts closely with aromatase; in the absence of AIPB, there is a significant amount of free aromatase. Given the regulation of estradiol synthesis by AIPB, changes in its expression may represent an early predictor of increased risk of developing breast cancer in premenopausal women. In addition, increasing AIPB expression may reduce estradiol synthesis in postmenopausal women.

MATERIALS AND METHODS

Resource for breast tissues.

We obtained human breast cancer tissue samples from a single surgical practice submitted by a single surgeon at the Anderson Cancer Institute at the Memorial Hospital in Savannah, GA. Study patients had core needle biopsy-proven invasive ductal carcinoma at least 1 cm in diameter and had consented to participate in the study through the standard process for an Institutional Review Board-approved study (IRB no. 2011.09.08), and its extension was approved on 7 July 2015. Tumorigenic and nontumorigenic tissues were definitively identified during surgery and were procured under ultrasound guidance using a 14-gauge disposable Bard biopsy gun (Bard, Tempe, AZ). The tissues were immediately transferred on ice and stored in liquid nitrogen, if not immediately processed for organelle fractionation or activity assay. The tissue specimens were placed on ice, and then the organelle fractions containing ER and the mitochondria from the epithelial cells were separated following a well-developed procedure (31, 32) that was partially modified.

Cell culture, transfection, and fractionation.

Human nontumorigenic MCF-12A (ATCC CRL-10782) and tumorigenic T-47D (ATCC HTB-133), MCF-7 (ATCC HTB-22), or MDA-MB-231 (ATCC HTB-26) breast cells were grown at a density of about 10⁵ cells/cm² and cultured as monolayers in Dulbecco's modified Eagle's medium (DMEM) supplemented with fetal bovine serum (FBS) (10% [vol/vol]), glutamine (2 mM), nonessential amino acids (1%), and penicillin-streptomycin (100×) under a humidified atmosphere of 5% CO₂ at 37°C. The medium was replaced with fresh medium every 2 days. MCF-12A cells were grown in DMEM–F-12 medium (1:1) containing 5% equine serum (ES), 20 ng/ml epidermal growth factor, 0.5 μg/ml hydrocortisone, 0.1 μg/ml cholera toxin, hygromycin (InvivoGen; catalog no. ant-hg-1, 100 mg/ml; lot no. HCG-38-02A) and 10 μg/ml insulin. Mouse Leydig (MA-10) cells (ATCC CRL-3050), African green monkey kidney COS-1 cells (ATCC CRL-1650), and human embryonic kidney HEK-293 cells (ATCC CRL-1573) were maintained followed same procedure as described before (33). All procedures were followed as recommended by ATCC. The room temperature was 24°C unless otherwise stated. All the restriction enzymes were purchased from New England Biolabs. Unless otherwise mentioned, all chemicals and media were purchased from Sigma, Fisher, or VWR.

For AIPB overexpression and activity determination, MCF-12A, MCF-7, and T-47D cells were plated at a density of 30,000 cells per well in a 6-well plate 18 h before transfection. The cells were washed first with phosphate-buffered saline (PBS) and then with 3% serum 16 h after transfections and supplemented with medium containing appropriate antibiotics and 10% FBS. To determine the expression levels, the transfected cells were collected after 48 h, washed with PBS, and lysed with 1× sample buffer. For activity determination, the cells were washed with PBS and gently collected after being scraped from the plate with PBS. After centrifugation at 3,000 rpm at 4°C, ER fractions were isolated, and estradiol was determined by radioimmunoassay (RIA) following the manufacturer’s procedure (estradiol [E2] double-antibody RIA kit; catalog no. SKU07138102; MP Biomedicals, OH). Isolation, fractionation, and purification of ER, mitochondrion-associated-ER membrane (MAM), and mitochondria were performed following our previously published procedure (21, 34). Cell growth/viability was determined following the previously described procedure (35) (VWR; catalog no. PAG8080 and lot no. 0000476331 with dispensed lot no. 0000456728). In brief, MCF-7 cells were plated in a six-well plate at an initial density of 30,000 cells per well for 24 h at 37°C in 5% CO₂. The media were changed to 3% serum containing 10, 25, 50, and 100 nM α-zeranol for 12, 18, 24, 36, 48, 72, and 96 h. The growth of stable AIPB-MCF-7 cells was determined in a similar fashion in the presence of 50 mM α-zeranol or 100 μg/ml of doxycycline for 24, 48, and 72 h. The density of cell growth was determined colorimetrically at 560 nm following 3 h incubation using FlexStation 5 (Molecular Devices, Sunnyvale, CA).

LC-MS/MS analysis.

SDS-PAGE-stained protein bands were excised, destained, reduced with dithiothreitol (DTT) (Roche Applied Science), alkylated with iodoacetamide (Sigma), digested with trypsin (Promega sequencing grade modified; Promega, Madison, WI) overnight and processed for mass-spectrometric analysis as previously described (36). All mass spectrometry results were obtained by LC-MS/MS using a Waters QTOF Premier mass spectrometer with a nanoAcquity ultrahigh-performance liquid chromatography (UHPLC) system. The UHPLC was equipped with a Waters trapping column (5-μm Symmetry C₁₈ column with the dimensions 180 μm by 20 mm) and a Waters analytical column (3-μm Atlantis dC₁₈ column with the dimensions 75 μm by 150 mm) and used 0.1% formic acid in water (solvent A) and 0.1% formic acid in acetonitrile (solvent B). A 5-μl portion of extract was injected and trapped for 3 min at a flow rate of 10 μl/min with 99% solvent A. Peptides were eluted with a linear gradient from 2% solvent B to 45% solvent B in 40 min. MS and MS/MS spectra were acquired using data-dependent acquisition. Peak lists were generated with ProteinLynx Global Server version 2.2.5 using mass tolerances of 0.1 Da for MS spectra and 0.2 Da for MS/MS spectra. Mascot MS/MS Ion Search (version 2.3) was used for database searching of the NCBInr nonredundant protein database (version from 19 October 2014), which contained 51,471,198 sequences. Consideration was allowed for two missed cleavages for trypsin (K and R) with carbamidomethylated cysteine and possible oxidation of methionine.

Cloning of AIPB.

5′ and 3′ RACE was used to clone the full-length AIPB cDNA from nontumorigenic human breast tissue. We first started with a generic 3′-primer called the adapter primer (AP) (GGC CAC GCG TCG ACT AGT ACT TTT TTT TTT TTT TTT T; Invitrogen/Life Technology) and a prepared first cDNA strand from total RNA. Using the first cDNA strand as the template, we amplified the second strand using a primer (AIPB11; CTG GAG GTC GTG GTG GAC CAG CCC ATG GAG AGG CTC) and an adapter primer, which generated a 600-bp fragment. Next, using two primers, AIPB12 (AGCT AGATCT ACC CTG GAG GTC GTG GTG GAC CAG CCC ATG GAG AGG CTC) and AIPB13 (AGCTA GAATTC TCA ACA CCT GGC TTC AGA GGC AGG), the amplified PCR product was cloned into the SP6 vector at the BglII and EcoRI sites. The sequencing result showed a completely new cDNA sequence, with a stop codon at 361 bp, suggesting the presence of an open reading frame further upstream. We then proceeded to 5′ RACE with the cDNA prepared from the same breast RNA. The cDNA was tail labeled with dCTPs and was amplified using a 5′-abridged anchor primer (AAP; GGC CAC GCG TCG ACT AGT ACG GGI IGG GII GGG IIG) and AIPB16, generating an additional 450-bp fragment. In the next step, full-length cDNA was cloned in the same SP6 in pCMV-flag vector (Stratagene, CA).

Western blot analysis and sources of antibodies.

Protein (12.5 μg) was separated by 15% SDS-PAGE and transferred to a polyvinylidene difluoride (PVDF) membrane (Millipore, Billerica, MA). Most of the antibodies were purchased from Abcam or Santa Cruz Biotechnology, and dilutions were performed by following their instructions. The membrane was blocked with 3% nonfat dry milk for 45 min, probed overnight with the primary antibodies, and then incubated with the peroxide-conjugated goat anti-rabbit IgG or anti-mouse IgG (Pierce). Signals were developed with a chemiluminescent reagent (Pierce). AIPB protein has 42 (∼20%) amino acid identity with the cholesterol trafficker (CT) protein; thus, we used a CT antibody to detect AIPB. CT is not expressed in breast tissue (18) and has no cross-reactivity with any other proteins present in breast. We have used our own VDAC2 antibody for mitochondrion-related experiments (21). Our VDAC2 antibody does not cross-react with VDAC1 (21). To confirm that CT is not expressed in the breast, we performed Western blotting with the C-terminus-specific CT antibody (Abcam; catalog no. ab133657 and lot no. GR97564). Antibodies to the following proteins were used for different experiments: aromatase (Abcam; catalog no. ab35604, lot no. GR323558-1 and GR323557-1; also catalog no. ab18995 and lot no. GR3187757-14), calnexin (Abcam; catalog no. ab22595 and lot no. GR190877), COX IV (Abcam; catalog no. ab14744 and lot no. GR192963-1), and GRP78 (Abcam; catalog no. ab21685 and lot no. 189602-1 and 189602-2).

AIPB activity in knockdown cells.

The Dharmacon/GE program was used to design siRNA sequences for all knockdown experiments. The program identified four siRNA oligonucleotides, where the first two were predicted to have 98% and 98.3% accuracy but the other two sequences had only 96% and 93% predicted accuracy. The first (sense, 5′GGGAGGAGGCCAUGCAGAAUU3′, and antisense, 5′UUCUGCAUGGCCUCCUCCCUU3′) and second (sense, 5′CACCUAGCACGUGGAUUAUU3′, and antisense, 5′UAAUCCACGUGCUAGGGUGUU3′) AIPB siRNAs were applied independently with 30 or 60 pmol Oligofectamine. The expression was determined by Western blotting. CT-specific siRNA sequences (A, sense, 5′CGUGGAUUAACCAGGUUCGtt3′, and antisense, 5′CGAACCGGUUAAUCCACGtg3′; B, sense, 5′CCAAACUUACGUGGCUACUtt3′, and antisense, 5′AGUAGCCACGUAAGUUUGGtc3′; C, sense, 5′GGAGAGUCAGCAGGACAAtt3′, and antisense, 5′AUUGUCCUGCUGACUCUCCtt3′) were used in MCF-12A and MA-10 cells to rule out the possibility of interference with CT in AIPB expression. Nonspecific or scrambled siRNA was a proprietary formulation of the manufacturer (Dharmacon).

To develop the biological activity assay, we used ¹⁴C-labeled testosterone and androstenedione as substrates following standard procedure (37). The metabolic conversion assay was carried out in a glass tube (VWR; 16 by 100 mm) in 50 mM potassium phosphate buffer (pH 7.4). Radiolabeled testosterone at a concentration of 0.5 UC (microcurie) was incubated with 100 μg of cell or tissue lysate as the source of enzyme (aromatase). To identify the role of AIPB, 30 pmol of siRNA1 and 30 pmol of siRNA2 were mixed together and transfected into the MCF-12A or T-47D cells using Oligofectamine in the absence of any serum. After 12 h, the medium was changed, and the knockdown experiment was continued for additional 36 h with medium containing serum and antibiotics. Next, the cells were washed with PBS twice, ER fractions were purified (34), and the lysate was prepared for metabolic conversion following a modified procedure (33). The reaction was initiated with 2 μg of cytochrome P450 and 2 mM NADPH following a standard procedure (37). For complete metabolic conversion, the reaction was chased with 10-fold cold testosterone and androstenedione independently. The reaction mixture was gently vortexed, and the reaction tubes, which were covered with Parafilm to avoid any evaporation loss during incubation, were incubated in a shaking water bath (∼40 rpm) at 37°C for 4 h.

Following incubation, 4 ml of ether-acetone (9:1) was added to each tube and gently vortexed to extract newly synthesized steroids. The tubes were allowed to sit at room temperature for about 10 min to separate the aqueous and organic layers. The upper, organic layer was gently collected without mixing the two layers using a Pasteur pipette and transferred to a new glass tube. The remaining reaction mixture was again subjected to organic solvent extraction. The extracted organic solvent layers were then mixed with 4 ml of basic water (0.01 N NaOH), vortexed gently, and allowed to remain for 15 min at room temperature to separate the layers. The upper organic solvent layer was collected in a fresh 5-ml glass tube (VWR International; 12 by 75 mm) and air dried. A cold testosterone-estradiol mixture was added to the completely dried reaction tubes at a final concentration of 0.1 mM resuspended in ethanol. The tubes were gently rolled to dissolve all the dried steroids, and 2 μl was counted in 2.5 ml of scintillation cocktail (Beckman Coulter, Brea, CA) in triplicate. Each sample was counted for 2 min, and 5,000 counts of each sample was then spotted on a silica-coated glass plate (20 by 20 cm; 60W F254S; Millipore, Billerica, MA). The silica plate was run in chloroform-ethyl acetate (3:1) for 1 h and dried in a 45°C air incubator before being exposed to a ³H screen. The radioactive thin-layer chromatography (TLC) method was changed to a radioimmunoassay (RIA) (estradiol double-antibody RIA kit; MP Biomedicals, OH; catalog no. 138102) because American Radiolabeled Chemicals stopped production of the substrate. Aromatase was also measured from the purified ER of the breast tissues or cells by ELISA (XpressBio, Frederick, MD; catalog no. XPEH2665). Each time, we used 12.5 μg of ER fractions for carrying out the metabolic reaction following the manufacturer’s procedure. In brief, following incubation of the tumorigenic (MCF-7 or T-47D) and nontumorigenic (MCF-12A) cells, the medium was removed, washed with 2×PBS, and then incubated with 2 mM HEPES (pH 7.0) for 30 min. After isolation of ER fractions, the protein concentrations were determined at every time point prior to activity measurement. For characterization of the steroids by gas chromatography-mass spectrometry (GC-MS), spots matching the autoradiogram were scraped from the silica plates and processed for characterization following previously described procedure (34). Spectra were collected in full scan mode with 70-eV ionization over the mass range of m/z 30 to 500 to facilitate the comparison of the MS spectra with the NIST/EPA/NIH NIST08 mass-spectral library.

Statistical analysis.

One-way analysis of variance (ANOVA) was performed to compare means of different samples. P values were determined to confirm any significant difference in the estradiol synthesis dependence on stimulation of cells with time and estrogen stimulator (38), and the P value was ≤0.05.

Production of cDNA and gDNA from MCF-12A and T-47D cells.

Total RNA from MCF-12A and T-47D cells was isolated using the Qiagen RNeasy minikit (Qiagen, Inc.; product ID 74104) according to the manufacturer’s protocol, and yield was quantified using a NanoDrop 2000 spectrophotometer (Thermo Fisher, Inc.). Total human ovarian RNA was purchased from Clontech, Inc. The samples were heated to 65°C for 5 min. Approximately 1 μg of purified total RNA from both cell lines was converted to cDNA utilizing LunaScript RT Supermix (New England Biolabs, Inc. [NEB]; product ID E3010) according to the manufacturer’s protocol. In addition to RNA isolation, genomic DNA was isolated from approximately 1 × 10⁶ cells for both cell lines using the Qiagen DNeasy blood and tissue kit (Qiagen, Inc.; product ID 69504).

Transmission electron microscopy.

Human breast tissue was gently washed with PBS and then transferred to 50-ml plastic disposable Corning tubes containing PBS. After centrifugation at 3,500 rpm (Beckman Allegra 22R and rotor F630) for 10 min, the tissue was fixed in 4% paraformaldehyde and 0.2% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.4, dehydrated with a graded ethanol series through 95%, and embedded in LR white resin. Thin sections of the nontumorigenic and tumorigenic breast tissue (75 nm thick) were cut with a diamond knife on a Leica EM UC6 ultramicrotome (Leica Microsystems, Bannockburn, IL) and collected on 200-mesh nickel grids. The sections were blocked in 0.1% bovine serum albumin (BSA) in PBS for 4 h at room temperature in a humidified atmosphere and incubated with aromatase (1:1,000) or CT (1:2,000) antibodies in 0.1% BSA overnight at 4°C. The sections were washed with PBS and floated on drops of anti-primary specific ultrasmall (<1.4 nm) Nanogold reagent (Nanoprobes, Yaphank, NY, USA) diluted 1:2,000 in 0.1% BSA in PBS for 2 to 4 h at room temperature. After PBS and deionized H₂O washes, the sections were incubated with HQ Silver (Nanoprobes) for 8 min for silver enhancement, followed by washing in deionized H₂O. Semiquantitative analysis was performed from the 40 best images obtained from several grids, with each image divided into 16 quadrants for counting the number of gold particles following a previously described procedure (34). To avoid any error, we counted each image five times (n = 5), and standard deviations (SD) were determined. Results were expressed as the number of gold particles per field of view and were calculated using the quantitation function of the Gatan Microscopy Suite software (Gatan Inc., Pleasanton, CA) as reported before (34).

PCR analysis.

PCR amplification of the predicted complete AIPB open reading frame and a portion of the human FEN1 gene was performed using OneTaq 2× master mix (NEB; product ID M0482). All primers were synthesized by either Integrated DNA Technologies, Inc., or Thermo Fisher, Inc. Amplification products from the AIPB PCRs (ELP1, sense, 5′GAT ACG CGT GCG GCC GCT CAT GGA ATA AAT TCT CCT CGA GAG3′, and antisense, 5′AAA GAA TTC GCG GCC GCG CCA CCA TGC TGC TAG CGA CAT TCA AGC3′) were analyzed by agarose gel electrophoresis (1.3% agarose) with 1× Tris-acetate-EDTA (TAE) buffer and 0.2 μg/ml ethidium bromide. Products from the FEN1 PCRs (ELP3, sense, 5′CCA GCT CTT CTT GGA ACC TG3′, and antisense, 5′CGC TCC TCA GAG AAC TGC TT3′) were analyzed by agarose gel electrophoresis (4% agarose) with 1× Tris-borate-EDTA (TBE) buffer and 0.2 μg/ml ethidium bromide. Gel images were captured with the Bio-Rad Gel Doc system (Bio-Rad, CA).

Generation of internal deletional mutants.

For generation of internal deletional mutants Δ13–17, Δ13–22, Δ13–27, and Δ13–32, the sense primers were JW2 (5′TGCGCTGGGAGCCGCAACAQTGAAGGGGCTGGCGCAACAG3′, for Δ13–17), JW3 (TGCGCTGGGAGCCTGGCGCAACAGCCTGTGATGGCCATC, for Δ13–22), JW4 (TGCGCTGGGAGCGTGATGGCCATCAGCCAGGAACTGAAC, for Δ13–27), and JW5 (TGCGCTGGGAGCCAGGAACTGAACCGGAGGGCCCTGGGG, for Δ13–32) and the antisense primer was DC1 (5′AGC TAG AAT TCT CAT GGA ATA AAT TCT CCT CGA GAG AA3′, for initial amplification). The final construct was developed with sense linker primer JW1 (5′AGC TGG ATC CAT GCT GCT AGC GAC ATT CAA GCT GTG CGC TGG GAG C3′) and antisense primer DC1 using each of the PCR products as a template. Next, all the PCR products were gel purified and digested with BamHI and EcoRI restriction enzymes followed by subcloning into pCMV-Flag (Stratagene, CA) as described previously (39). Taq (Pfu) polymerase was purchased from Thermo Fisher, and the deoxynucleoside triphosphates (dNTPs) were from Roche Molecular Sciences (Foster City, CA).

Generation of stable AIPB-expressing clones.

The cloned AIPB cDNA sequence was utilized for constructing a stable, doxycycline-inducible AIPB in MCF-7 cells. For this, primers SM2KZ (5′-AAAGAATTCGCGGCCGCGCCACCATGCTGCTAGCGACATTCAAGC-3′) and SM1 (5′-GATACGCGTGCGGCCGCTCATGGAATAAATTCTCCTCGAGAG-3′) were used for In-Fusion cloning (TaKaRa Bio, Inc.; catalog no. 638909, lot no. 1706009A) into the NotI site of the pTETOne inducible expression system (TaKaRa Bio, Inc.; catalog no. 634303) to create vector pTETOneAIPB. In the absence of a selectable marker in pTETOne, plasmid pCpGfree-vitroHmcs (InvivoGen, Inc.) was utilized for cotransfection and providing hygromycin resistance. The plasmid pTETOneAIPB was linearized with PvuI and pCpGfree-vitroHMCS (InvivoGen; catalog no. pCpGfree-vitroHmcs G2 TDS 09E11-MM) with PacI.

Next, 250 ng of each digested vector was transfected into MCF-7 cells using Lipofectamine LTX according to the manufacturer’s protocol (Thermo Fisher; catalog no. 15338100, lot no. 1879097). As a control, MCF-7 cells were transfected with an empty vector without pTEToneAIPB. Stable clones were selected and maintained at 100 μg/ml in Hygromycin Gold (InvivoGen; catalog no. ant-hg-1 [stock, 100 mg/ml] and lot no. HCG-38-02A). For doxycycline induction experiments, cells were harvested after 24 h of induction (1 μg/ml) by trypsinization, and total RNA was isolated using the RNeasy minikit (Qiagen, Inc.) including the supplemental addition of RNase-free DNase I in the RNA preparation according to the manufacturer’s protocol. The final RNA concentration was determined using the NanoDrop 2000 spectrophotometer. For Western blotting and RIA, the AIPB stable cells and MCF-7 cells were transfected with hrGFP expression plasmid (200 ng/well) (available from Edward Perkins’ lab) in a six-well plate in the absence of antibiotics. After 16 h, the medium was changed and supplemented with serum and hygromycin in stable cells and 1× penicillin-streptomycin in MCF-7 cells. The cells were collected after 42 h. The GFP antibody was used at a 1:2,000 dilution (Thermo Fisher; catalog no. PAI-980A; kindly provided by Robert Visalli of our department) in Western blotting.

For generation of stable cell lines in nonsteroidogenic cells, AIPB plasmid was subcloned into pCDNA3.1 (Invitrogen, CA) and then COS-1 cells were transfected with the plasmid using Lipofectamine (Invitrogen) following the previous procedure (40). Single clones were generated 48 h after transfection by limiting dilution into a selection medium containing 600 μg/ml G418 (Geneticin; Invitrogen, CA) following a previously published procedure (33, 40, 41). Individual clones were then transferred to 24-well plates (VWR) for propagation and later transferred to 6-cm plates. Individual clones were examined for AIPB expression by Western blotting, as described in “Western blot analysis and sources of antibodies” above.

Co-IP analysis.

Specific antibodies were preincubated with protein A-Sepharose CL 4B (0.5 μg/μl; GE, NY) in 100 μl of 1× co-IP buffer (1% Triton X-100, 200 mM NaCl, and 0.5% sodium deoxycholate). After mixing for 2 h at 4°C, the beads were washed with 1× co-IP buffer five times and then incubated again with rabbit IgG control antibody (Sigma) for 1 h. After another wash series, freshly isolated ER pellets (25 mg/sample) were resuspended with ice-cold lysis buffer (20 mM Tris HCl [pH 8.0], 137 mM NaCl, 10% glycerol, 1% Triton X-100, 2 mM EDTA) at 4°C for 15 min. Insoluble material was removed by ultracentrifugation (30 min at 100,000 × g). The supernatants were incubated overnight at 4°C in the presence of antibodies prebound to protein A-Sepharose beads. After washing with 1× co-IP buffer and 10 mM HEPES (pH 7.4), the protein A-Sepharose pellets were resuspended and vortexed with 100 mM glycine (pH 3.0) for 10 s. After addition of a pretitrated volume of 1.0 M Tris (pH 9.5), and the beads were separated by centrifugation at 2,000 × g for 2 min. The supernatants (immune complexes) were analyzed by Western blotting.

In vivo chemical cross-linking.

Chemical cross-linking is a process of chemically joining two or more molecules by a covalent bond. Cross-linking reagents are called cross-linkers, where the specific molecules contain two or more reactive ends capable of chemically attaching to specific functional groups (primary amines, sulfhydryls, etc.) on proteins or other molecules. To study direct interaction of proteins, in vivo cross-linking was performed with minor modification of a previously described procedure (21). MCF-12A and T-47D cells (5 × 10⁶) were grown in tissue cultures dishes, washed twice with PBS at room temperature, and then collected by gentle scraping. Next, the cells were incubated with the cross-linker BS3, which was solubilized in water to a working concentration of 50 mM. BS3 is amine reactive in that its N-hydroxysulfosuccinimide (NHS) esters at each end react with primary amines to form stable amide bonds. The BS3 cross-linker has an 8-atom spacer (11.3 Å) and not a cell membrane-permeative cross-linker. After incubating the cells with 0.5, 1.0, 2.0, and 5 mM cross-linker at 37°C for 1 h in a rotating shaker, the reaction was quenched by addition of 1 M Tris (pH 7.4) to a final concentration of 50 mM for an additional 15 min at 4°C. To avoid any endogenous protease activity, we immediately added a protease inhibitor mixture (Pierce), and the incubation continued for an additional 15 min at room temperature. The cross-linked cells were collected by centrifugation at 3,000 rpm and resuspended in 10 mM HEPES (pH 7.4). The bulk in vivo cross-linking was performed with 2 mM BS3 following our standard procedure with minor modification (21, 34). For in vitro chemical cross-linking, the isolated ER fractions were incubated with various concentrations of BS3 solubilized in water in a similar fashion described above. The reactions were terminated either by transferring them on ice or by the addition of 10 μl of 1.0 M Tris buffer (pH 7.4) depending on the experimental requirement.

AIPB-aromatase interaction and activity determination.

MCF-12A, T47-D, and MCF-7 cell lysis was prepared by a freeze fracture method by incubating cells three times at −80°C followed by transfer to room temperature and mild vortexing. Next, CT (1:1,000) and aromatase (1:1,000) antibodies were added independently to 100 ng of cell lysate and incubated for 30 min at room temperature. As a control, we also incubated MCF-7 cell lysate with rat IgG under identical conditions, and estradiol activity was determined as described above.

Data availability.

The 624-bp cDNA sequence coding for the AIPB protein has been deposited in GenBank under accession no. MT920320 (under the name SAM). Primary data and other supporting materials created or accumulated in the course of the work will be shared with other researchers upon reasonable request and will be made available appropriately through material transfer agreement between the two institutions. Reuse and redistribution of our reagents will be regulated by the Mercer University policy in order to protect privacy and confidentiality concerns, as well to respect any proprietary or intellectual property rights. Legal offices will be consulted to address any concerns, if necessary. Terms of use will include proper attribution to the principal investigator and authors along with disclaimers of liability related to any use or distribution of the research materials.