Phosphorylation of human placental aromatase CYP19A1

Abstract

Aromatase CYP19A1 catalyzes the synthesis of estrogens in endocrine, reproductive and central nervous systems. Higher levels of 17β-estradiol (E2) are associated with malignancies and diseases of the breast, ovary and endometrium, while low E2 levels increase the risk for osteoporosis, cardiovascular diseases and cognitive disorders. E2, the transcriptional activator of the estrogen receptors, is also known to be involved in non-genomic signaling as a neurotransmitter/neuromodulator, with recent evidence for rapid estrogen synthesis (RES) within the synaptic terminal. Although regulation of brain aromatase activity by phosphorylation/dephosphorylation has been suggested, it remains obscure in the endocrine and reproductive systems. RES and overabundance of estrogens could stimulate the genomic and non-genomic signaling pathways, and genotoxic effects of estrogen metabolites. Here, by utilizing biochemical, cellular, mass spectrometric, and structural data we unequivocally demonstrate phosphorylation of human placental aromatase and regulation of its activity. We report that human aromatase has multiple phosphorylation sites, some of which are consistently detectable. Phosphorylation of the residue Y361 at the reductase-coupling interface significantly elevates aromatase activity. Other sites include the active site residue S478 and several at the membrane interface. We present the evidence that two histidine residues are phosphorylated. Furthermore, oxidation of two proline residues near the active site may have implications in regulation. Taken together, the results demonstrate that aromatase activity is regulated by phosphorylation and possibly other post-translational modifications. Protein level regulation of aromatase activity not only represents a paradigm shift in estrogen-mediated biology, it could also explain unresolved clinical questions such as aromatase inhibitor resistance.

Introduction

The enzyme aromatase (AROM; CYP19A1), an integral membrane protein of the endoplasmic reticulum, is synonymous with estrogen because it is the only one known to catalyze the synthesis of estrogens. In the presence of its electron transfer partner cytochrome P450 reductase (CPR), AROM converts androstenedione (A) and testosterone (T) to estrone (E1) and 17β-estradiol (E2), respectively, by aromatization of the steroid ‘A’ ring (C1-C5, C10) [1,2], as shown in Figure 1. In human, AROM is the only member of the CYP19 family and the CYP19A sub-family. Since all ten exons of the CYP19A1 gene splice onto a common 3′-splice junction upstream of the ATG site, the coding region and the encoded protein are the same [2,3]. It is, thus, the same protein everywhere in the human organs and tissues, such as ovary, breast, endometrium, placenta, and the central nervous system (CNS). Higher levels of E2 are associated with malignancies and diseases of the breast, ovary, and endometrium, while low E2 levels increase the risk for osteoporosis, cardiovascular disease, and cognitive disorders. About 70% of all breast cancer cases are estrogen-dependent [4,5] and AROM inhibitors (AIs) are the drugs of choice in endocrine therapy for estrogen-dependent post-menopausal breast cancers. As an essential female reproductive hormone, E2 is the transcriptional activator of the estrogen receptors. However, the genotoxic effect of E1 and E2 metabolites causing mutation by DNA adduct formation, as schematically illustrated in Figure 1, has also been suggested as a possible mechanism for tumorigenesis [5–12].

Figure 1. Estrogen biosynthesis and signaling pathways.

Aromatase (AROM; CYP19A1) catalyzes the conversion of androstenedione (A) and testosterone (T) to estrone (E1) and 17β-estradiol (E2), respectively. Some of the other enzymes involved in E1⇔E2 inter-conversion (activation-deactivation) and genotoxic product formation are also shown. Three possible pathways of estrogen action are schematically depicted. Two examples of E1/E2-derived genotoxic mutations are provided. Each pathway represents a simplified summary of the current understanding of the area from its vast literature, some of which are referenced in the introduction. Ade: adenine, Gua: guanine.

AROM in the temporal cortex and hippocampus is localized in neurons and astrocytes [13–15]. Regulation of AROM activity by phosphorylation (RAAP) in the auditory cortex of songbirds, rapid estrogen synthesis (RES) (the so-called ‘E2 burst’) at the synapse, and non-genomic E2 signaling as a neurosteroid via plasma membrane receptors (Figure 1) have been widely reported [16–20]. Postsynaptic signaling by the brain estrogen has also been linked, sometimes sex-specifically, to amyotrophic lateral sclerosis (ALS), Alzheimer’s disease, nociception, and epileptic seizures [20–30]. Nevertheless, RAAP is yet to be unequivocally established, even after 60 years of discovery of the enzyme. RES and possible non-genomic effects of E2 have rarely been described in the context of the human endocrine-reproductive axis, and hence remain incompatible with the current paradigms of estrogen-dependent malignancies and disorders. Indeed, RES and the resulting overabundance of E1/E2 coupled with genomic, non-genomic and genotoxic signaling (Figure 1) could rapidly stimulate estrogen-responsive malignancies and disorders in the endocrine organs, reproductive tissues and CNS. Contrary to the brain AROM scenario, there are only a handful of reports to date on phosphorylation of AROM in cell lines [31–34].

For decades, a stumbling block to understanding the phosphorylation phenomenon has been the absence of AROM’s 3-D atomic structure. Recent advances made in this area provide unprecedented insights into this challenging enzyme and shed new light on its function [35–44]. Here, by combining biochemical, cellular, high-resolution tandem mass spectrometry and structural biology approaches, we present a comprehensive description of AROM phosphorylation at the atomic level, and demonstrate unequivocally that the human enzyme purified from normal term placenta is phosphorylated at multiple sites. We show that phosphorylation/dephosphorylation at a strategic location can directly influence the enzyme’s catalytic activity.

Materials and methods

Generation of phospho-Y361-specific AROM antibody (PY361Ab) and WB analysis

The residue Y361 is located in the K-helix on its exposed face within the heme-proximal cavity [36], at the AROM–CPR coupling interface. We have already established that Y361 is critically important for catalysis [36]. Two polyclonal antibodies PY361Ab (phospho-Y361-specific Ab) and nPY361Ab (nonphospho-Y361-specific Ab) were generated (by GenScript, NJ) against the peptides ³⁵⁰IQKLKVMENFI-(±phospho)-Y³⁶¹ESMR³⁶⁵, respectively. Both were affinity-purified and verified for specificity by ELISA (data not shown). The detectability of these antibodies was determined to be ~1–5 ng when used at 1: 1000 dilution. Phosphorylated and non/dephosphorylated AROM samples were analyzed by SDS–PAGE and transferred to nitrocellulose membranes. The blots were blocked in TBST (25 mM Tris (pH 7.4), 135 mM NaCl, 3 mM KCl and 0.1% Tween 20) solution supplemented with 3% nonfat milk (w/v) for 30 min at room temperature, and subsequently washed with TBST. Antibodies nPY361Ab and PY361Ab were diluted 1 : 1000 in TBST supplemented with 1% nonfat milk (w/v) and incubated with the membrane at 4°C for a minimum of 2 h and maximum of 16 h. The membranes were then washed extensively with TBST and incubated with horseradish peroxidase-linked goat-anti-rabbit IgG diluted 1 : 1000 in TBST supplemented with 1% nonfat dry milk (w/v) for a minimum of 2 h. After washing three times with TBST, the immunoblot signals were visualized by enhanced chemiluminescence detection using the Clarity Western ECL Substrate Kit (Bio-Rad, Hercules, CA) following the manufacturer’s protocol.

Phosphorylation of AROM

Purified human AROMs, from term placenta (pAROM) [35] and recombinantly expressed in E. coli (rAROM) [36], were phosphorylated with Src kinase (SrcK) (Signal Chem, BC, Canada, Cat# S19-10G). The target site Y361 and the corresponding kinase (SrcK) were selected as per NetPhosK 2.0 sever prediction and previous reports [31,32]. Western blot (WB) analysis was performed with both PY361 and nPY361Ab’s with freshly purified pAROM as the control. Two rAROM mutants Y361F and Y361D were also used. Dephosphorylation of the Y361-phosphorylated p/rAROM was done by PTPN1 (Signal Chem, BC, Canada) as per manufacturer’s protocol.

Mass spectrometry (MS)

Gel bands of rAROM corresponding to 55 kDa monomer and 110 kDa dimer were excised and washed three times with acetonitrile; the final wash contained ammonium bicarbonate. For the purified pAROM samples, the solutions were directly subjected to proteolysis. Trypsin digestion was then carried out (1 : 10 molar ratio of trypsin to protein) by incubation at 37°C for 16 h. The non-alkylated cysteine digest samples were then analyzed by LC–MS/MS using either a Q-Exactive Plus or an Orbitrap Fusion mass spectrometer equipped with a Waters nanoACQUITY ultra-performance liquid chromatography (UPLC) system using a Waters Symmetry C18 180 μm by 20 mm trap column and a 1.7 μm (75 μm inner diameter by 250 mm) nanoACQUITY UPLC column (35°C) for peptide separation. Trapping was done at 15 μl/min with 99% buffer A (100% water, 0.1% formic acid) for 1 min. Peptide separation was performed at 300 nL/min with buffer A and buffer B (100% acetonitrile, 0.1% formic acid) over a linear gradient. High-Energy collisional dissociation was utilized to fragment peptide ions via data-dependent acquisition. Mass spectral data were processed with Mascot Distiller, utilizing the high-resolution profile peak-picking algorithm. Protein searches were conducted against the homo sapiens SwissProt protein database (20 240 sequences) using Mascot Search Engine (Matrix Science, LLC, Boston, MA; v. 2.6.0). Mascot search parameters included: parent peptide ion tolerance of 10.0 ppm, peptide fragment ion mass tolerance of 0.020 Da, strict trypsin fragments (enzyme cleavage after the C terminus of K or R, but not if it is followed by P), variable modification of phospho (S, T, Y, and H), oxidation (M), and propioamidation (C). Mascot peptide scores were given for each phosphorylation site identified, indicating how well the ions were fragmented and how well it matched with theoretical fragmentation patterns. Assigned phosphorylated peptides and sites of phosphorylation with ions scores that indicate identity or extensive homology (P < 0.05) were manually validated through investigating the MS/MS fragmentation pattern of the phosphopeptides. Those that fall slightly under this criterion were also verified manually and with additional orthogonal/structural supporting evidence (see results). Ion scores and mass accuracy in ppm values were also given for each identified phosphorylation site. Mass accuracy values were calculated according to the equation [(Theoretical mass – Observed Mass)/(Theoretical Mass)].

TiO₂ enrichment

Phosphorylated peptides were enriched using a TiO₂ (titanium dioxide) TopTips column (TiO₂; GlySCi, Columbia, MD) with a slightly modified manufacturer protocol. Briefly, the manufacturer protocol was utilized with the addition of 70 mM l-glutamic acid in the loading buffer (65% acetonitrile, 2% trifluoroacetic acid). Bound phosphopeptides on the TiO₂ resin were washed with 65% acetonitrile, 2% trifluoroacetic acid, and eluted with 2% ammonium hydroxide solution in water at pH 12. Both enriched phosphopeptides and flow-through (i.e. un-enriched) peptides fractions were analyzed by LC–MS/MS as described above.

Measurement of AROM-specific activity

The specific activities of AROM before and after phosphorylation/dephosphorylation were determined by the release of tritiated water following the 1β-³H, 4-¹⁴C androstenedione to E1 aromatization reaction [45,46]. An amount of 1 μl of each sample was reconstituted in 1,2-didodecanoyl-sn-glycero-3-phosphocholine and AROM activity was measured in triplicate or quadruplicate with excess purified rat CPR and 5 mM NADPH. Any inconsistent data were re-measured and outliers were excluded from the averaging.

Analysis of phosphorylation in MCF-7 cells

A breast cancer cell line MCF-7a, stably transfected with AROM cDNA [47], was used to establish Y361 phosphorylation. AROM expression and activity in regular MCF-7 cells are low and not easily detectable [48,49]. The cells were cultured as previously described [37] and detached from the tissue culture flask according to standard protocol. For E2 treatment, 1 nM of E2 was added to the media after 24 h of incubation and the media was refreshed three more times after every 48 h. The cells were harvested by trypsinization, transferred to a 1.5 ml eppendorf microcentrifuge tube and spun at 13 200 rpm for 1 min. The supernatant was aspirated without disturbing the pellet. The cells were resuspended in 75–200 μl AROM buffer (20% glycerol in 10 mM phosphate buffer solution pH 7.4 having 0.5 μM androstenedione, 0.1 mM EDTA, 0.5 mM DTT, and 1 mM n-dodecyl-β-d-maltoside) containing protease inhibitors. The optimal resuspension volume for 1.5 × 10⁶ live cells/ml was 100 μl. The solution was thoroughly vortexed and placed in ice for 30 min. The cell lysate was prepared by removing the debris by centrifugation. SDS-PAGE, WB, and AROM-specific activity analysis were performed as per protocols described above.

Preadsorption

To validate specific epitope recognition by antibodies in WBs in impure samples such as cell and tissue extracts, and to eliminate false-positive recognition due to cross-reactivity, PY361Ab and nPY361Ab were preadsorbed with their respective epitope peptides used for their generation. Using a spin column, Affi-Gel 10 resin (Bio-Rad, Hercules, CA) was incubated with 3–5 μg excess peptide overnight at 4°C on a rotisserie. After overnight incubation, 10 μl of ethanolamine (pH 8) was added to bind unreacted sites. The mixture was incubated for an additional hour at room temperature. The column was then spun down and washed five times with 0.1 M sodium-acetate (containing 0.5 M NaCl) and 0.1 M NaHCO₃, once with 400 μl PBS, and six times with 10 mM Tris pH 7.5. The PY361Ab was added to the column at the same dilution as used for the WB and incubated for 2 h at room temperature. After 2 h, the column was spun down and the flow through containing the unbound antibodies was diluted in 1% nonfat milk and used for western blotting.

X-ray crystallographic analysis and molecular modeling

Extensive details of crystallographic analysis have been previously published [35–38,44]. The data deposited with the PDB bank and the methods described therein were used to generate models and electron density maps. No new structure determination was undertaken for this work. The software used for molecular modeling and illustrations was UCSF Chimera [50] version 1.12 running on a MacOS platform.

Results

Src-kinase catalyzed phosphorylation of Y361

WB analyses were performed with the phospho-Y361-specific and the nonphospho-Y361-specific antibodies PY361Ab and nPY361Ab, respectively, using freshly purified human placental pAROM [35] as the control (see methods for details). Two mutants Y361F and Y361D of the recombinant enzyme rAROM [36] were also used. The results are shown in Figure 2. In all numbered lanes in Figure 2A–D, 1 μg of the pure enzyme was used, except for the left panel of Figure 2A where the amounts are specified. Typically, pAROM on the denaturing SDS–PAGE runs mostly as a 55 kDa monomer (1mer), 5–10% as 110kDa dimer (2mer), and sometimes at high protein concentrations as 165kDa trimer (3mer) (Figure 2A, left panel). The recombinant rAROM is missing the 39 N-terminal transmembrane amino acids, but has the added 10 hydrophilic residues at the N terminus, and four histidine residues at the C terminus [36]. Thus, the rAROM monomer is shorter than pAROM by 25 residues and runs at slightly below pAROM at 55 kDa in the WB with nPY361Ab that labels the normal, nonphosphorylated monomeric AROM (Figure 2A, middle panel). The antibody nPY361Ab is also able to detect weakly the dimer bands at ~110 kDa. As expected, the WB with PY361Ab is unable to recognize either band of p/rAROM (Figure 2A, right panel).

Figure 2. SDS–PAGE and WB of p/rAROM, pre- and post Y361 phosphorylation.

(A) Left panel: SDS–PAGE of purified pAROM; middle and right panels: WB of p/rAROM; 1mer: monomer, 2mer: dimer, 3mer: trimer. (B) SDS–PAGE of pAROM, Y361F and WT rAROM pre- (lanes 1, 3 and 6, respectively) and post- (lanes 2, 4 and 7, respectively) phosphorylation. (C) WB with PY361Ab of pAROM, Y361F and WT rAROM post-phosphorylation by SrcK (lanes 2, 4 and 6, respectively); lanes 1, 3 and 5: p/rAROM controls; lane 7: post-PTPN1 treatment of lane 6 sample. (D) WB with nPY361Ab of pre- and post-phosphorylation pAROM (lanes 1 and 2, respectively), and post-phosphorylation rAROM WT and Y361F (lanes 3 and 4, respectively) labeling nonphosphorylated p/rAROM as 55 kDa bands.

Figure 2B is a denaturing SDS–PAGE before and after phosphorylation of pAROM, rAROM Y361F and the wild-type (WT). Routinely, a stronger than normal (as in Figure 2A, left panel) dimer band is visible on phosphorylation (lanes 2 and 7); this band is absent for the control (lanes 1, 3, and 6) and rAROM Y361F mutant (lane 4). This suggests that Y361-phoshorylated AROM is more prone to denaturation, and has a more exposed epitope when aggregated as a dimer under the SDS–PAGE conditions, than the Y361-nonphosphorylated AROM (i.e. the native pAROM and the WT rAROM). SrcK (with the GST tag) runs at 83 kDa (SignalChem, Cat# S19-10G) and is not detectable under our experimental condition. In Figure 2C, PY361Ab recognizes the p/rAROM 55 kDa monomer bands weakly, but the 110 kDa dimer strongly (more exposed epitope) on phosphorylation (lanes 2 and 6), and no AROM bands for the pre-phosphorylation p/rAROM controls (lanes 1, 3, and 5) and post-phosphorylation Y361F (lane 4). Lane 7 represents the result of dephosphorylation by PTPN1 of lane 6 phosphorylated rAROM. Lower molecular mass bands in lane 4 are non-specific, possibly due to cross-reactions to SrcK contaminants in the absence of a specific epitope. Figure 2D is a WB with nPY361Ab of pre- and post-phosphorylation pAROM (lanes 1 and 2) and post-phosphorylation rAROM WT and Y361F (lanes 3 and 4) recognizing the nonphosphorylated portions of p/rAROM as 55 kDa bands.

Effect of Y361 phosphorylation on enzyme activity of pAROM, WT rAROM, and Y361F, Y361D mutants

The specific activities were measured on six pAROM and rAROM samples each from six independent phosphorylation experiments. The results are shown in Figure 3 and Table 1. We observe on the average 1.5 fold (range 1.3 to 1.6) and 2.4 fold (range 1.3 to 3.7) enhancement in specific activities of pAROM (Figure 3A) and rAROM (Figure 3B), respectively, on phosphorylation with SrcK in comparison with the nonphosphorylated samples under the same conditions as the controls. Additionally, specific activities for Y361D and Y361F mutants phosphorylated identically with SrcK exhibited no significant difference from the controls, averaged over three and two experiments, respectively. Although these results show a substantial jump in activity on phosphorylation of Y361, the WB data suggest that large portions of the samples were not phosphorylated (Figure 2D). Furthermore, when phosphorylated AROM is de-phosphorylated with the phosphatase PTPN1, 21% loss of activity was observed averaged over two experiments (Figure 3C and Supplementary Table S1).

Figure 3. Assay of p/rAROM specific activities pre- and post-Y361 phosphorylation.

(A) pAROM and (B) WT rAROM, shown are averages 5 experiments. rAROM Y361D/F data are averages of three experiments. (C) Decrease in specific activity after phosphatase PTPN1 treatment of phosphorylated rAROM, averaged over two experiments; symbols are ** P < 0.005, *** P < 0.0005, **** P < 0.00005, ns: not significant.

Table 1.

Effect of phosphorylation by SrcK on specific activity (SA) of purified AROM

	pAROM			rAROM WT			Y361D			Y361F
Expt. no.	SA control sample	SA phospho sample	Fold increase in SA	SA control sample	SA phospho sample	Fold increase in SA	SA control sample	SA phospho sample	Fold increase in SA	SA control sample	SA phospho sample	Fold increase in SA
1	26.5	33.5	1.3	11.7	23.2	2.0	21.1	11.7	0.6	9.8	7.3	0.7
2	43.4	63.0	1.5	11.3	22.7	2.0	13.8	10.5	0.8	5.8	4.9	0.8
3	43.0	62.4	1.5	17.7	22.7	1.3	10.5	10.5	1.0	-	-	-
4	41.7	62.4	1.5	7.4	27.6	3.7	-	-	-	-	-	-
5	8.5	13.0	1.5	4.8	17.6	3.7	-	-	-	-	-	-
6	5.5	9.0	1.6	10.1	18.6	1.8	-	-	-	-	-	-
Avg ± SD			1.5 ± 0.1			2.4 ± 1.0			0.8 ± 0.2			0.8 ± 0.2

SA is in nmole/min/mg; Avg, average; SD, standard deviation.

All experiments are performed in triplicate and the averages are entered.

The Y361F mutant itself has greatly reduced expression levels and its purification is far from routine, whereas Y361D behaves like the WT and expresses well. When compared with the WT enzyme, Y361F has roughly 90% reduced activity, while the Y361D activity appears to be similar to the WT enzyme (Supplementary Table S2).

Y361 phosphorylation in a MCF-7 breast cancer cell line

The cell lysates were used to perform WB analysis with PY361Ab/nPY361Ab and enzyme activity measurement. The results are shown in Figure 4. In Figure 4A, the anti-nonphosphoY361 antibody nPY361Ab recognized the pAROM control (as in Figure 2A) at the monomeric 55 kDa, none in MCF-7 lysate, and a weak 55 kDa nonphosphorylated AROM monomer band, as well as a stronger band at 40 kDa and one at even lower MW, which may be degradation products of AROM. All AROM-specific bands, including the pAROM control, disappeared in the WB shown in Figure 4B when nPY361Ab was preadsorbed with the nonphospho-Y361 epitope peptide. In Figure 4C PY361Ab recognized neither the pAROM control, nor any band in the MCF-7 lanes, but strongly labeled the 55 kDa monomer in the MCF-7a lysate, lane 6, indicating Y361 phosphorylation, and none as the 110 kDa dimer band. SDS-induced denaturation and/or aggregation of AROM molecules in the presence of other cellular components, and the resulting exposure of the epitope could possibly be different from those of the isolated AROM in the cell-free medium. This would result in altered sensitivity to the antibody of different molecular species of AROM as demonstrated by the differences in the labeling of the monomeric and dimeric bands between WBs of Figures 2 and 4. The AROM-specific band is no longer labeled by PY361Ab when preabsorbed with the phospho-Y361 epitope peptide, as shown in Figure 4D. AROM activities in MCF-7a whole cell lysate was significantly higher those in MCF-7 and MCF-7a treated with E2 that exhibited background level activities (Figure 4E and Supplementary Table S3). Treatment of MCF-7a cells with E2 and T repeatedly showed weak 55 kDa AROM band and stronger lower MW bands in WBs with both nPY361Ab and PY361Ab (data not shown) suggesting instability/degradation of AROM.

Figure 4. Western blots (WB) with nPY361Ab and PY361Ab of MCF-7 and MCF-7a whole cell lysates.

Lanes in (A–D) are 1: MW markers, 2: 100 ng purified pAROM control; 3, 4, 5: MCF-7 lysate; 6: MCF-7a lysate. (A) nPY361Ab recognizes pAROM (same as in Figure 2A), none in MCF-7 lysate, and a weak 55 kDa nonphosphorylated AROM monomer band and stronger lower MW bands, which may be degradation products of AROM. All AROM-specific bands, including the pAROM control, disappear in WB (B) when nPY361Ab is preadsorbed with the epitope peptide. (C) PY361Ab recognizes neither the pAROM control, nor any band in the MCF-7 lanes, but strongly labels the 55 kDa monomer in the MCF-7a lysate, lane 6, indicating Y361 phosphorylation, and probably no 110 kDa dimer band. (D) The AROM-specific band in (C) is no longer labeled by PY361Ab when it is preabsorbed with the phosphopeptide epitope. (E) AROM activity in MCF-7a whole cell lysate is significantly higher those in MCF-7 and MCF-7a treated with E2 that exhibit background level activities. ** indicates P < 0.005.

Mass spectrometric determination of phosphorylation sites

Mass spectrometric analysis of purified p/rAROM by LC–MS/MS protocol was conducted multiple times, and with various samples, yielding consistent results. These p/rAROM samples were prepared exactly the same way as the enzymes had routinely been purified from normal term human placenta and recombinant expression cells, and prepared for the past biochemical and crystallographic studies [35–38,40,44]. The most reliable LC–MS data and MS/MS fragmentation assignments of the modification sites were obtained by analyzing the pAROM phosphopeptides enriched with TiO₂ treatment. The flow-through of the TiO₂ column (see methods) was also analyzed as un-enriched pAROM. The results presented in Table 2 demonstrate that the naturally occurring placental enzyme is phosphorylated at multiple sites. At least three of the highest-scoring LC–MS phosphopeptides corresponding to the sites T462, T162, and H475/S478 were identified in all MS analyses performed to date, even in the absence of enrichment. By TiO₂ enrichment, however, the same LC–MS phosphopeptides were detected with higher scores and lower ppm values, leading to the identification of as many as 19 phosphorylation sites in naturally occurring human pAROM (Table 2). Again, phosphopeptides containing the residues H475/S478, T462, and T162 topped the list suggesting that these sites have highest degrees of phosphorylation. As expected, the peptides from the first 64 N-terminal residues, including the 40 N-terminal transmembrane helix residues, were never observed, simply because the first R/K appears at position 64 in the sequence. Presence of the complete N terminus was confirmed by Edman degradation sequencing of the first 10 N-terminal residues (data not shown).

Table 2.

LC–MS/MS identification and assignment of pAROM phosphorylated peptides enriched with TiO₂

Residue	Score1	ppm	Phosphopeptide	Times detected	Comments
S478 (H475)	67	0.7	I474-HDL-phosS-LHPDETK485	2 (En2)	S478 is a catalytic residue in the active site. R192 nearby. H475 is the assigned phosphorylation site in un-enriched sample. R79 in proximity
T462	64	0.7	phosT462-LQGQCVESIQK473	8 (En & FT3)	At lipid interface. K324 is nearby to neutralize/stabilize charge.
T162	61	−0.8	M160V-phosT-VCAESLK169	6 (En & FT)	Residue with a hint of electron density for the phosphate moiety. Close to AROM–CPR coupling interface. R159 to neutralize/stabilize.
S211	61	−0.4	I206PLDE-phosS-AIVVK216	2 (En)	Consistent with structure; R193 nearby
Y386	57	−0.2	A377LEDDVIDG-phosY-PVKK390	3 (En)	Deepest site within lipid bilayer; K99 nearby
Y361	54	2.6	V355MENFI-phosY-ESMR365	2 (En)	Y361 OH interacts with S182 side chain from the symmetry-related molecule in the crystal structure. Positive charge of K448 side chain is nearby, but currently salt-bridges to E357; K448 +ve charge, ~5 Å away, is poised to interact with the Y361 phosphate.
S470	51	−0.5	T462LQGQCVE-phosS-IQK473	2 (En)	Inside lipid layer. Consistent with structure. No +ve charge nearby. H-bond with Q466.
T179	45	−2.5	L175EEV-phosT-NES182GYVDVLTLLR192	1 (En)	Consistent with structure; no positively charged residue nearby
S267/ T268	43	0.3	R265I-phosS-TEEKLEECMDFAT280ELILAEK287	2 (En)	Either is possible from the structural point of view. R264 and 265 are available for stabilization/charge neutralization.
T493	43	0.8	N486MLEMIF-phosT-PR495	2 (En)	Near the exposed C terminal. K461 stands in close proximity.
T143	38	0.1	phosT143-TRPFFMK150	1 (En)	Highly exposed region; K142, the next residue on the helix, is poised to interact with the phosphate moiety.
S101	37	−1.3	S100-phosS-SMFHIMK108	2 (En)	Inside lipid layer. S101 is most exposed of 3 serine residues. K99 and R425 side chains are nearby.
T392	37	−0.3	G391-phosT-NIILNIGR400	1 (En)	Inside lipid. Structurally feasible; K389 is nearby.
T414	36	−1.0	P410NEF-phosT-LENFAK420	1 (En)	Inside lipid layer. Structurally feasible; K409 is nearby.
Y184	35	0.8	L175EEVTNESG-phosY-VDVLTLLR192	1 (En)	Near lipid interface. Consistent with structure. No intramolecular +ve charge in close proximity; K354/K448 from symmetry-related molecule
Y366	31	−1.7	phosY366-QPVVDLVMR375	2 (En)	Inside lipid bilayer; R403 nearby
T201	31	0.1	V194MLDTSN-phosT-LFLR205	1 (En)	Consistent with structure. R205 side chain is in close proximity.
S72	28	−0.6	F65LWMGIG-phosS-ACNYYNR79	1 (En)	Inside lipid layer; phosphorylation obstructed by phosphate bound to H475

¹For the highest-scoring phosphopeptide, in case of multiple detections.

²En: enriched sample.

³FT: Flow-through (un-enriched).

The biochemical results presented above in Figure 2 establish that Y361 is susceptible to phosphorylation and dephosphorylation by SrcK and PTPN1, respectively. In order to further validate this data beyond WB recognition of phosphorylated Y361, the 55 kDa and the 110 kDa SDS–PAGE bands of WT rAROM before and after phosphorylation (Figure 2B, right panel) were subjected to MS analysis. The LC–MS data on the post-phosphorylation 55 kDa monomeric band revealed the additional Y361-containing phosphopeptide on SrcK phosphorylation (Supplementary Figure S1A,B). The corresponding MS/MS fragmentation spectrum, although weak, is consistent with Y361 phosphorylation (Supplementary Figure S2A). Interestingly, the same LC–MS phosphopeptide was also obtained with TiO₂ enrichment from pAROM even though it was not treated with kinase, suggesting that Y361 is naturally phosphorylated in pAROM (Table 2). However, the MS/MS fragmentation data of the phosphopeptide is strong but equivocal as to whether Y361 or S363 is the phosphorylation site (Supplementary Figure S2B). Nevertheless, the location of S363 in the 3-D structure prevents it from being phosphorylated as it is totally unexposed and tightly packed in a hydrophobic pocket (side chain −OH making a hydrogen bond to backbone C = O of F359). It is highly unlikely that S363 could be phosphorylated without disrupting the integrity of the heme scaffold. The more likely scenario, therefore, is that Y361 in naturally occurring human pAROM is phosphorylated at levels too low to be detected by WB. Although the LC–MS data confirmed the identity of the 110 kDa band as AROM, no phosphopeptide was identified by analyzing this dimeric band, probably due to the low protein level. The coverage was only 33% as opposed to 68% coverage obtained from the 55 kDa band. Low coverage could also result from fewer available cleavage sites due to non-specific aggregation of the denatured protein. More denatured protein under the SDS–PAGE condition, and a more exposed epitope are consistent with the stronger 110 kDa (2mer) WB band on phosphorylation of Y361, in Figure 2C.

Lastly, MS analysis was also performed on an untreated and unenriched pAROM sample, yielding the same top-scoring LC–MS phosphopeptides with residues T462, T162, and H475/S478 (Supplementary Figure S3A). The MS/MS fragmentation analysis of the phosphopeptide I⁴⁷⁴HDLSLHPDETK⁴⁸⁵ identified H475 as the phosphorylation site in un-enriched pAROM with 99% probability (Supplementary Figure S3B). However, the fragmentation data of the same LC–MS phosphopeptide from pAROM after TiO₂-enrichment determined S478 to be the site and not H475 (Supplementary Figure S3C). Thus, MS/MS fragmentation analyses of the same phosphopeptide I⁴⁷⁴HDLSLHPDETK⁴⁸⁵ from two different pAROM samples identified two different residues, H475 and S478, as the phosphorylation sites. The fragmentation spectra of some of the other manually validated assignments for phosphorylated peptides listed in Table 2 are provided in Supplementary Figure S4. For each of these phosphopetides, the corresponding nonphosphorylated peptide from each AROM sample was analyzed and identified (data not shown).

Effect of phosphate ion on AROM activity

Phosphate buffer has a demonstrable positive effect on the stability and activity of AROM. All extraction, purification, and crystallization of p/rAROM were performed in phosphate buffer, pH ~ 7.5. When the phosphate ions were removed by exchanging with Tris buffer of similar pH’s, we observed a significant reduction in the specific activities — 33% for pAROM and 67% for rAROM (Supplementary Table S4). This data is a small representative sample of decades of observation on extraction, purification AROM from microsomal fractions of human placenta.

The structures of human p/rAROM provide a molecular basis of interaction of phosphate ions with the enzyme. In all p/rAROM crystal structures determined and examined to date (>30), we identified two conserved phosphate ion binding sites, both involving histidine side chains H475 and H111 shown in Figure 5 with the electron density maps calculated from the crystal structure of pAROM at 2.75 Å resolution, PDB ID: 3S79 [37]. While H109/H111 is located near the AROM–CPR interaction interface where Y361 is located, the location of H475 is in close proximity to the active site, inside the lipid bilayer. At both sites, the phosphate oxygen atom makes a short hydrogen bond forming contact with N1 of the histidine imidazole ring, while the D side-chain carboxylate makes a hydrogen bond with N3 by donating the proton. The geometries of the triads ${\text{PO}}_4^{3-}\text{-H111-D381}$ and ${\text{PO}}_4^{3-}\text{-H475-D371}$ (Figure 5A,B), in which the D carboxylate and the phosphate oxygen atoms are nearly in the plane of the H imidazole ring, are similar to those observed for histidine phosphorylation sites, such as the one shown in Figure 6A,B illustrating ADP and ${\text{PO}}_4^{3-}$ bound structures of human nucleoside diphosphate kinase (NM23-H1) (PDB ID’s: 2HVD and 3L7U, respectively) [51,52]. Figure 6C shows the structure of the phospho-H intermediate formed in the homologous drosophila NM23-H1 (PDB ID: 1NSQ) [53]. Both E and D carboxylates are found in histidine kinases for phosphorylation either at the N1 or the N3 position of the imidazole ring. Figure 6D is the active site of EnvZ, a bacterial histidine kinase/phosphotransferase, in complex with a non-hydrolyzable ATP analog AMPPNP (PDB ID: 4KP4) [54]. Similarity of the geometries of the ${\text{PO}}_4^{3-}\text{-H-D/E}$ triads (co-planarity of the D/E carboxylate and the phosphate oxygen atoms to the H imidazole ring) in Figures 5 and 6 suggest that both H111 and H475 would be phosphorylated at the N1 position. Phosphorylation of H111, however, could not be detected by LC–MS/MS. NM23-H1 and EnvZ are auto-phosphorylated at histidine N1 and N3 positions, respectively, with ATP as the phosphate donor [54,55]. It is unknown if ${\text{PO}}_4^{3-}$ could by itself act as a nucleophile and auto-phosphorylate by acid-base catalysis.

Figure 5. Conserved phosphate ion binding sites at locations critical to catalysis.

Shown are the phosphate ion-omitted (2Fo – Fc) electron density maps contoured at 2σ for the 2.75 Å resolution Parom structure (PDB ID: 3S79) around the phosphate ions bound (A) to H475 in close proximity to the active site access channel, and inside lipid bilayer; D371 carboxylate and ${\text{PO}}_4^{3-}$ oxygen atoms make hydrogen bonds with H475 N-atoms roughly in a planar geometry with the imidazole ring, and (B) between H111 and H109 near the AROM–CPR coupling interface; similarly, D381 carboxylate and ${\text{PO}}_4^{3-}$ oxygen atoms make hydrogen bonds with N-atoms of the H111 imidazole ring in a planar geometry. The presence of positively charged R79 in (A), and K108 and N110 in (B) near the phosphate moiety is to be noted.

Figure 6. Phosphate binding at the catalytic histidine auto-phosphorylation motif in the active site of nucleoside diphosphate kinase (NDPK) NM23-H1 and phosphotransfer signaling kinase EnvZ.

Structures of (A) ADP complex and (B) phosphate-bound form of human NM23-H1 (PDB IDs: 2HVD and 3L7U, respectively). (C) phospho-histidine intermediate in the homologous drosophila enzyme (PDB ID: 1NSQ) as the first step of phosphate transfer catalyzed by NDPK. (D) Active site of EnvZ that transfers the γ-phosphate from ATP to another amino acid is shown in complex with a non-hydrolyzable ATP analog (PDB ID: 4KP4). Thus, the geometry of phosphate binding, ${\text{PO}}_4^{3-}\text{-H475}/\text{H111-D}$ in AROM, and the presence of positively charged R and K shown in Figure 4 are similar to those in phosphate-binding motifs in (A–D). Residue E of the phosphate-binding motif in (A–C) is replaced by a D in (D).

Structural mapping and perspective on phosphorylation

In Figure 7A–C, we mapped all the Y, T and S phosphorylation sites identified by LC–MS/MS in purified pAROM, listed in Table 2. Phosphorylated side chains were modeled by building phosphate esters of the hydroxyl groups on the experimentally observed orientations. Minimal adjustment of the side-chain conformations, if any, was necessary to achieve this, suggesting that phosphorylation at these sites were consistent with the 3-D atomic structure on the enzyme, with one exception. S72 phosphorylation is obstructed by the presence of the phosphate ion bound to H475 (Figure 7A,B). The H sites, H475 and H111 (Figure 7A,C), are left as phosphate ion binding sites, as experimentally observed in the human AROM crystal structures. Another important revelation is the presence of R and/or K side chains in the immediate vicinity of the phosphoryl moiety for most of the phosphorylation sites, as described in Table 2. These positively charged side chains appear to neutralize and/or stabilize the negatively charged phophoryl groups. The exceptions are residues T179, Y184, and S470 that lack positive charges in close proximity. The sites that are located either within the lipid bilayer or at the lipid-protein interface are S72, S101, Y184, Y366, Y386, T392, T414, T462, S470, and H475 (Figure 7A–C), some of which may have roles in catalysis, transmembrane CPR coupling, substrate entry/product release, maintenance of the active site architecture as a membrane-embedded molecule, or preservation of the water channels [35,44]. Residues at or near the CPR coupling interface are S101, H111, T143, T162, Y361, and Y386. These residues may have direct coupling interaction with the redox partner CPR, especially with its FMN domain [42,56,57]. Sites T179, S201, S211, S267, and S493 are also in the cytoplasmic side of the molecule.

Figure 7. Mapping of phosphorylation sites on membrane-embedded full-length pAROM.

(A) All Y, T, S phosphorylation sites are shown as phosphoryl side chains. The H-sites, H475 and the proposed H111, are shown bound to phosphate ions, as observed in the crystal structure. Modeling of S72 phosphorylation is obstructed by phosphate bound to H475. All other 17 phosphorylation sites are structurally feasible and easily modeled on the crystal structure. Also shown are R and K side chains present near all phosphorylation sites except T179 and S470. The block arrow delineates passage of the substrate from lipid bilayer to the active site through the access channel. The lipids in the bilayer are depicted in thin wire model for clarity. Dotted red and blue boxes in (A) outline the regions shown in (B) and (C), respectively, and color-matched arrows indicate the approximate viewing direction. Close-up views in (B) and (C) of the phosphorylation sites surrounding the catalytic cleft and the CPR-coupling interface are: (B) Phosphate ion-bound H475 located inside the lipid bilayer, in close proximity to the active site residues including phospho-S478; the substrate access channel indicated by the arrow. (C) Several phosphorylation sites, including Y361 and the proposed H111, shown with the bound phosphate ion, line the CPR-coupling interface, at the heme-proximal surface.

Oxidation of proline residues

In addition to phosphorylation, the LC–MS/MS analysis also identified two proline oxidation sites — P138 and P481. The MS/MS fragmentation spectra are shown in Supplementary Figure S5A,B. An increase in 16 Da for each oxidized proline residue was attributed either to the opening of the five membered ring via the formation of γ-glutamic semialdehyde, or generation of hydroxyproline [58–62]. Although the implications of oxidation of these two proline residues, if any, remain to be determined, their locations in the structure are quite interesting. P138 is situated at the N terminus of helix aC (Figure 8A), at the end of αB′-loop-αB-loop, that could serve as the ‘lid’ to the active site, opposite to the substrate access channel [35]. The residue P481 is located at the C terminus of β8–β9 antiparallel strands, the gateway to the substrate access channel. The β8–β9 antiparallel strands also host the two catalytically important phosphorylation sites H475 and S478 (Figure 8B).

Figure 8. Proline oxidation sites P138 and P481 in the crystal structure of AROM.

(A) P138 (circled in red) is located at the N terminus of helix αC, at the end of αB′-loop-αB-loop, that could serve as the ‘lid’ to the active site, opposite to the substrate access channel. Oxidation of P138 and possible opening of the five-membered ring via the formation of γ-glutamic semialdehyde could provide more flexibility of the ‘lid’, thereby creating another egress/entry pathway from/to the active site. (B) P481 is located at the C terminus of β8–β9 antiparallel strands, gateway to the substrate access channel. The β8–β9 antiparallel strands also host two catalytically important phosphorylation sites H475 and S478. Oxidation of P481 and possible opening of the five-membered ring could provide more flexibility of the region providing less stringent access to the active site, and perhaps aiding in the H475 to S478 phosphate transfer process as well.

Discussion

AROM from the human placenta has been studied for more than 50 years. Yet its phosphorylation has remained obscure, thereby confounding interpretation of experimental and clinical data. Although evidence for up- and down-regulation of AROM activity, and reports of RAAP and RES as well as non-genomic E2 signaling at synapses have been widely published, neither direct evidence of AROM phosphorylation nor identification of the phosphorylation sites has emerged. In the endocrine-reproductive systems, AROM phosphorylation has remained virtually unknown, barring a handful of reports of phosphorylation in breast cancer cell lines [31–33] and recently a computational analysis [63]. High throughput studies in cell lines occasionally stumbled upon phosphopeptides from AROM [64], e.g. the same phospho-T162 containing AROM peptide was one of the 1761 phosphopeptides from 491 proteins identified by a LC–MS/MS analysis of HeLa cells [65].

The data presented here unequivocally demonstrate AROM phosphorylation in human placenta and map the phosphorylation sites in the 3-D atomic structure. We also show that phosphorylation of Y361 boosts enzyme activity in vitro. LC–MS/MS analysis on the pAROM and WT rAROM for the first time revealed that AROM is phosphorylated at multiple sites. Furthermore, by TiO₂ enrichment we confirm that pAROM, the enzyme extracted from normal human placenta, has as many as 19 phosphorylation sites, of which T462, T162, H475/S478, and Y361 are major and reproducibly detectable. Additionally, data from the enrichment experiment suggest that the human placental AROM is innately phosphorylated at Y361, albeit at levels lower than easily detectable by WB. Phosphorylation of S478, located in the active site access channel (Figure 7B), is also highly significant since previous studies already implicated S478 in catalysis [35,36,38,66]. Mutations S478A and S478T resulted in reduced V_max in comparison with the WT enzyme in both in-cell and microsomal assays [66]. Therefore, S478 phosphorylation could be directly involved in the regulation of catalysis. Secondly, H475 and S478 are located on a short antiparallel strands β8–β9, in close proximity to each other (Figures 7B and 8). Modeling of S478 phosphorylation shows no steric conflict and the possibility that these two side chains can get even closer by conformational adjustments. It is, thus, plausible that H475 functions as a phosphate transfer site for S478.

By designing and utilizing a pair of anti- phospho- and nonphospho-Y361-specific Ab’s, we show by WB that SrcK can phosphorylate this residue and validate the result by LC–MS/MS. Furthermore, we also demonstrate by WB that Y361 is phosphorylated in a MCF-7 breast cancer cell line. The SDS–PAGE and WBs of Y361-phosphorylated p/rAROM in the cell-free media show the monomer and dimer bands (Figure 2), while the WBs of MCF-7a labeled the enzyme as a monomer (Figure 4). This data suggests that denaturation and aggregation patterns of the isolated enzyme and the enzyme in the presence of other molecules in the cellular environment, under the SDS denaturing conditions, are different. However, it could simply be an experimental artifact and may not have any physiological meaning.

Surprisingly, no phosphorylation of any site was detected in more than 30 p/rAROM crystal structures determined to date, with possibly one exception. As shown in Supplementary Figure S6, the appearance of elongation of the T162 electron density at the OH end at low contour levels provided the only hint of phosphorylation in the crystal structure. It is likely that the purified enzyme is a cohort of heterogeneously phosphorylated populations, and only the nonphosphorylated population crystallizes. Variation in the basal level-specific activity of freshly purified pAROM, previously reported to range roughly between 30 and 100 nmol/min/mg [35,37,46], is probably a reflection of the mixed population of the enzyme phosphorylated at varying sites to varying degrees.

Interestingly, examination of the phosphorylation sites in the crystal structure revealed that all Y/T/S/H phosphorylation sites are on the solvent-accessible surface and modeled easily as phosphorylated side chains without any significant steric hindrance. This probably indicates that phosphorylation leaves the 3-D structure virtually in tact. Nonetheless, it is plausible that some phosphorylation sites are associated with local conformational perturbations, which could result in a conformationally heterogeneous population of AROM molecules, with varying degrees of overall stability. The in vivo regulatory function, dictated among many factors by the presence and/or absence of certain kinases and phosphatases, may necessitate phosphorylation/dephosphorylation of only a critical few of these residues. Some of the sites may, indeed, have purely structural implications, e.g. stability of membrane integration, flexibility of the substrate access-product egress channels, and operation of the CPR coupling interface. Nevertheless, phosphorylation and dephosphorylation at these sites may still influence activity through long-range effects.

While the implications of some of the major phosphorylation sites, such as H475, S478, and Y361, are seemingly apparent, others are not. Roles of two prominent sites, T162 located at the protein exterior near the CPR-binding interface, and T462 at the lipid–protein interface, are yet to be elucidated. We have shown that Y361 phosphorylation is activating. The data suggest that phosphorylation of H475 (and possibly H111) could also be activity enhancing. Is phosphorylation at T162, T462, and S478 or any other residue inhibitory or activating? Why does then purified human AROM appear to be fully active? Would dephosphorylation/phosphorylation at these sites elevate its activity even further to a hyperactive AROM, or inactivate the enzyme altogether? Does deactivation result from destabilization/denaturation upon phosphorylation? Most importantly, what is the molecular/chemical mechanism by which the phosphorylation sites collectively accelerate or decelerate the rate of catalysis? Does phosphorylation influence AROM–CPR coupling and/or the CPR-to-AROM electron transfer rate? A primary and direct consequence of phosphorylation and/or dephosphorylation could be instantaneous modulation of the electromotive force that drives the transfer of electrons from the FMN moiety of CPR to the heme-Fe of AROM, as we previously postulated for Y361 phosphorylation [41]. Our future work would seek answers to these critical outstanding questions.

Histidine phosphorylation is also an intriguing finding that raises the obvious question — is it a phosphate transfer site? Phosphorylation of histidine is abundant in bacterial proteome and well studied. However, in the mammalian systems, although histidine phosphorylation is recognized as a legitimate regulatory mechanism, it is yet to be extensively investigated [67]. One reason is that histidine phosphorylation is rather unstable and hence difficult to establish, either by WB or MS. Under the acidic conditions of MS analysis, the stability of phospho-histidine is further reduced [67–69]. Surprisingly, LC–MS/MS of un-enriched pAROM was able to detect H475 phosphorylation while TiO₂ enrichment identified S478 as the phosphorylation site instead. Nevertheless, the peptide I⁴⁷⁴-HDLSLHPDETK⁴⁸⁵ containing these two residues is one of the highest-ranking LC–MS phosphoteptides in all samples analyzed to date (Table 2 and Supplementary Figures S1 and S3). Comparison of the geometries of the phosphate ion binding sites of AROM and human NM23-H1 leads to the proposition that, like NM23-H1, the two H-sites in AROM could possibly be auto-phoshorylated (Figures 5 and 6). The question is — do ATP molecules bind at these sites for auto-phosphorylation, like NM23-H1 in Figure 6? To date there has been no report of ATP binding to AROM; we would like to explore this possibility in the near future.

The structural data that H111 and H475 are two conserved phosphate ion binding site is interesting in this regard. That phosphate buffer enhances AROM activity is in agreement with a 1989 report that also showed loss of activity in the presence of phosphatase [70]. Can phosphate ions themselves auto-phosphorylate the histidines? In that case, the histidine side chain, in the presence of the aspartate as the general base, could act as the nucleophile attacking P of the phosphate moiety. A protonated arginine/lysine side chain serves as the general acid for donating the proton to P–O⁻ promoting the formation of P–N bond and release of a water molecule, as schematically depicted in Supplementary Figure S7. For H475 phosphorylation, D371 is the general base activating H475. R79, the positively charged side chain that stabilizes phosphate ion binding, could act as the general acid by donating a proton to P–O⁻, and accepting an electron pair therefrom. N75 nearby could have some role in the mechanism. For H111 phosphorylation, D381 is the general base and H109 could serve as the general acid while N110 could also play a role in stabilizing the phosphate ion binding and phosphorylation. Alternatively, it is possible that AROM requires one or more specific histidine kinase(s) and ATP for phosphorylating the histidines. Although H111 phosphorylation is yet to be proven, our data show that both H475 and S478 are phosphorylated, but in different populations of molecules. This supports the hypothesis that H475 phosphate, besides being itself unstable under the MS conditions, is transferable to S478. Moreover, it has been suggested in the gas phase peptides not only undergo loss of phosphorylation, but also phosphate group rearrangement [71].

The existence of positively charged residues in proximity to all but three phosphorylation sites is noteworthy as well. Positive charges are presumably there for neutralization and stabilization of the negatively charged phosphate moieties. Besides, they are also likely to play important roles in the phosphorylation process, either enzymatic or auto-phosphorylation. However, not all R/K side chains are necessarily at ideal geometry, in terms of distance and relative orientation of the phosphate moiety, for optimal interactions. Nonetheless, each site has enough flexibility for adjustment of the geometry for stronger stabilizing interactions between the R/K and the phosphate charges.

Another surprising finding of the MS analysis is proline oxidation. The implications of oxidation of P138 and P481 described above are not yet understood. It is not even clear if oxidation of these proline residues has any regulatory role or is somehow related to the regulation by phosphorylation. Amino acid oxidation in proteins could result from oxidative stress and enzymatic metabolic pathways [59,60]. However, the locations of these two proline residues, shown in Figure 8, are intriguing in the sense that both are near the active site. Oxidation of P138 and possible opening of the five-membered ring via the formation of γ-glutamic semialdehyde could provide more flexibility of the ‘lid’ αC (Figure 8A), thereby creating another egress/entry pathway from/to the active site. Similarly, oxidation of P481 (Figure 8B) and opening of the five-membered ring could relax the restrictive nature of the access to the active site maintained by the β8–β9 gateway [35] and perhaps aiding the putative H475-to-S478 phosphate transfer process as well.

Does AROM phosphorylation have anything to do with AI resistance? Could RAAP be responsible for AI resistance, at least in part? When used as therapeutic agents in a neoadjuvant or metastatic setting, about 25 to 50% of estrogen-dependent breast cancer patients do not respond to AIs, and ~10–15% patients relapse within 5 years [72]. The mechanism of AI resistance has been and continues to be an unsolved mystery [73,74]. It has been suggested that ineffective and/or insufficient inhibition of intratumoral AROM by AI [73,75], and increased AROM activity [76] could be the mechanisms for AI resistance. Given the identification of multiple phosphorylation sites in the close neighborhood of the active site and the redox partner-binding interface, it is plausible that phosphorylation of some of these residues could interfere with AI binding. By the same token, enhancement in the activity owing to phosphorylation in a certain population of molecules could overwhelm the inhibitory effects of AIs in others. Recognition of the role of each critical site could lead to specific diagnostic and preventive agents for reversing the detrimental effects of hyperactivity of the enzyme and/or ineffective inhibitors.

In conclusion, we have presented here the evidence of RAAP–RES in human placental AROM raising the possibility of the local overabundance of estrogens in the endocrine-reproductive systems. The same mechanism of protein level regulation is probably at work in other organs and tissues, and is responsible for RES, e.g. bursts of estrogen at the synaptic terminals. The system of multiple and alternate trigger points of activation and deactivation may afford the enzyme the capability of acute and rapid control of the activity in a wide range of tissue and cellular environments. While phosphorylation at some residues could serve to elevate the catalytic activity, it may reduce the activity at other sites. The cumulative effects of phosphorylation and/or dephosphorylation occurring at few selected sites could be the essence of regulation and RAAP–RES under various physiological situations. Contrary to the current paradigms that restrict the role of E1/E2 only to estrogen receptor-mediated signaling, sudden overabundance of estrogens could manifest itself by triggering additional signaling pathways, such as non-genomic signaling via plasma membrane receptors and genotoxic effects of its metabolites forming DNA adducts and causing mutations. It remains to be seen if RAAP–RES is markedly aberrant in human diseases and can help explain otherwise inexplicable clinical data. The presence of active AROM and its phosphorylation status could thus be the critically important biomarkers, independent of the estrogen receptors, in diagnosis and treatment of estrogen-dependent diseases.

Abbreviations

AIs

AROM inhibitors

ALS

amyotrophic lateral sclerosis

CNS

central nervous system

CPR

cytochrome P450 reductase

RAAP

regulation of AROM activity by phosphorylation

RES

rapid estrogen synthesis

UPLC

ultra performance liquid chromatography

WB

Western blot

WT

wild-type.