Aromatase in Human Bone Tissue



Peripheral aromatization of androgens exert estrogenic actions in many tissues. Recently in situ production of estrogens by aromatase was detected in human bone and cultured osteoblasts and has been proposed to participate in the maintenance of bone mass. We examined aromatase expression by immunohistochemistry and mRNA in situ hybridization in 16 cases of tibia (female 2 male, 14 female, 62 ± 5.2 years old) and quantified the level of aromatase mRNA in 28 cases of rib, femur, and lumbar vertebrae (16 male, 12 female, 58.0 ± 11.3 years old) by reverse transcriptase-polymerase chain reaction (RT-PCR) in order to study whether or not and in which cell types aromatase was expressed in human bone tissues. We also studied alternative use of multiple exons 1 of its gene and immunolocalization of type I 17β-hydroxysteroid dehydrogenase (HSD), which converts estrone produced by aromatase to estradiol. Strong aromatase immunoreactivity and mRNA hybridization as well as type I 17β-HSD immunoreactivity were detected in lining cells, osteoblasts, chondrocytes of articular cartilage, and adipocytes adjacent to bone trabeculae in all the cases examined. Amounts of aromatase mRNA varied greatly among the subjects (11.25 ± 9.77, 0.61–42.84 attomol/ng of total RNA). The amount of aromatase expression was not correlated with age or gender of the subjects but positively correlated with the degree of osteroporotic changes evaluated by radiological findings of lumbar vertebrae. Analysis of multiple exons 1 revealed that 1b or fibroblast type was predominantly (23/26) utilized as a promoter of aromatase gene expression. These results demonstrated that aromatase is expressed widely in human bone tissue and may play important roles in maintenance of human bone tissue.


BONE REMODELING IS REGULATED by systemic hormones and locally produced factors acting in concert to maintain bone mass. Among these hormones, estrogens have marked effects on bone metabolism.1 Various clinical and epidemiological studies have demonstrated that estrogen is one of the protective factors against osteoporosis, and estrogen deficiency is also one of the important factors in developing osteoporosis in postmenopausal women.1–4 Factors other than estrogen metabolism are involved but it is also true that not all postmenopausal women develop problems with osteoporosis despite decreased serum estrogen concentration.5 In addition, androgens have been known to affect bone mass.5 The ovary is the main contributor to circulating estradiol in premenopausal women but estrogen biosynthesis after menopause is mainly peripheral, through conversion of androstenedione or C19 steroids from the adrenal cortex and ovaries.6,7 This peripheral process takes place in skin,7 muscle,8 adipose tissue,9 and others, and this conversion is catalyzed by the enzyme complex cytochrome P450 aromatase.7–9 In sex-steroid dependent neoplasm, including breast, endometrial, and ovarian carcinoma, increased aromatase expression and/or activity in the tumor tissue has been demonstrated to be closely correlated with malignant phenotype10–12 despite the fact that there has been no consistent evidence of increased serum estrogen concentration in women with these tumors.13 In situ estrogen biosynthesis by aromatase has also been demonstrated to influence estrogen-related various neuronal and regional brain function.14

In bone tissue, in which both androgens and estrogens exert their effects, aromatase activity has been recently demonstrated by Schweikert et al.15 Nawata et al. also reported the presence of aromatase activity and mRNA in primary cultured osteoblast-like cells from normal human bones and osteoblast-like osteosarcoma cells.5 However, aromatase expression including its localization has not been studied in bone tissues. Therefore, we studied aromatase expression in 16 cases of tibia by immunohistochemistry and mRNA in situ hybridization, and quantified the level of aromatase mRNA in 28 cases of rib, femur, and lumbar vertebrae by reverse transcriptase polymerase chain reaction (RT-PCR) to examine whether or not and in which cell types aromatase is expressed in human bone tissue. We then correlated the level of aromatase mRNA with the degree of osteoporosis, determined by radiological findings of vertebrae, age, and gender of the subjects. Recently, expression of the aromatase gene involving alternative splicing of multiple forms of exon 1 has been demonstrated to contribute to tissue-specific expression of aromatase and its overexpression.16–18 Therefore, we examined exon 1 of the aromatase gene in 28 cases of rib and lumbar vertebrae by RT-PCR to determine whether alternative splicing is present or not and whether any specific exon 1 is correlated with aromatase overexpression. We also immunolocalized type I 17β-hydroxysteroid dehydrogenase (HSD), which converts estrone, the product of aromatase, to estradiol, a potent estrogen19 in 16 cases of tibia. We then correlated these findings to elucidate the biological significance of aromatase expression in human bone tissue.


Human bone tissue

Twenty-eight cases of human bone tissues stored at −80°C were retrieved from the tissue bank of the Department of Pathology, Tohoku University School of Medicine, Sendai, Japan. Ages (58.6 ± 11.3 years old) and gender (16 male, 12 female) and the sites (16 rib, 9 lumbar vertebrae, 3 femur) of the subjects examined are summarized in Table 1. Rib bones were obtained from open thoracic surgery or autopsy. Lumbar vertebrae were obtained from autopsy. Femur bones were obtained from amputation due to osteosarcoma or chondrosarcoma. Only parts of spongiosa had been removed from these specimens and cut approximately into 1.0 × 1.0 × 1.0 cm3 sections. They had been frozen in liquid nitrogen and stored at −80°C. This investigation was approved by the Ethics Committee on Human Study of Tohoku University School of Medicine.

original image

The degree of osteoporotic changes in these subjects was determined by retrospective review of radiological findings of lumbar and/or thoracic vertebrae, according to the criteria of primary osteoporosis of The Japan Bone Mineral Metabolism Society.20 They are summarized as follows; 0, normal; I, an increase in prominence of vertical situation; II, sparse vertical situation; III, loss of vertical situation; (+), the presence of collapse or compression fractures; (−), the absence of collapse or compression fractures. I (−) is radiologically defined as osteopenia. I (+), II, and III are radiologically defined as osteoporosis. Sixteen cases (2 male, 14 female, 56–78 [62 ± 5.21] years old) of nonpathologic tibia, fixed in 4% paraformaldehyde (pH 7.4) for 2 h at room temperature, decalcified in 10% EDTA for 72 h at room temperature, and embedded in paraffin were retrieved from osteopathology files of Department of Pathology, Tohoku University Hospital. These specimens were obtained from various orthopedic surgeries for degenerative or rheumatoid arthritis. None of the patients received corticosteroids prior to surgery.

Statistical analysis

We analyzed the correlation between sex and the amount of aromatase mRNA using Wilcoxon's rank-sum test, that between age and aromatase using Spearman's correlation coefficient and the comparison between degree of osteoporosis and aromatase using Dunnett's multiple comparison test. The level of significance was less than 0.05.


Rabbit immunoglobulin G antiaromatase antibody was prepared against the enzyme as purified from human placenta.21 Preparation of the antibody, immunoblotting, and immunohistochemical techniques have been described previously by the authors.21,22 Rabbit anti-17β-HSD type I was generated against human placental protein,19,23 and was kindly provided by Dr. Matti Poutanen and Dr. Reijo Vihko, Department of Clinical Chemistry, University of Oulu, Oulu, Finland.


Paraffin-embedded tissue blocks were cut into 2.5-μm sections and mounted on clean glass slides. Deparaffinized sections were placed in methanol containing 0.3% hydrogen peroxide for 3 minutes to inhibit endogenous peroxidase activity. The sections were washed in three changes of 0.01 mol/L phosphate-buffered saline (PBS) and incubated with normal goat serum for 30 minutes at room temperature. Immunohistochemical staining was carried out by a biotin-streptavidin-amplified method, using the Histofine immunostaining system (Nichirei, Tokyo, Japan). Details of immunohistochemical procedures were previously described by the authors.10,11,22 Control sections were incubated with normal rabbit serum or 0.01 mol/l PBS. No immunoreactivity was observed in these control sections. Positive control was normal cycling human ovary. Both aromatase and type 1 17β-HSD immunoreactivity was detected in membrana granulosa of dominant follicle. To characterize immunopositive cells, we performed following stains in adjacent tissue sections; immunostain of alkaline phosphatase using the monoclonal antibody against bone-specific alkaline phosphatase obtained from SAOS-2 human osteosarcoma cell line (kindly donated by Mitsubishi Kagaku Bio-Clinical Laboratories Inc., Tokyo, Japan)24 as a marker of osteoblast and histochemistry of tartarate-resistant acid phosphatase (TRAP) as a marker for osteoclast according to Barka and Anderson25 with some modification.26

DNA probes

The sequence of the 27-base aromatase oligonucleotide probe used for in situ hybridization analysis was as follows: GCG CAT GAC CAA GTC CAC GAC AGG CTG (847–873). The use of this DNA oligonucleotide probe in in situ hybridization analyses of aromatase mRNA in normal cycling human ovaries and endometrial cancer have been previously described by the authors.12,22 Sense oligonucleotide probes were used as negative controls. The probes were synthesized with a 3′ biotinylated tail (Brigati tail) (5′-probe-biotin-biotin-biotin-TAG-TAG-biotin-biotinbiotin-3′).27

In situ hybridization

In situ hybridization was performed with the MicroProbe staining system (Fisher Scientific, Pittsburgh, PA, U.S.A.) using manual capillary actions, with modification of methods previously published.24 Tissue sections (3 μm, applied to ProbeOn Plus slides, Fisher Scientific) were rapidly dewaxed, cleared with alcohol, rehydrated with a Tris-based buffer, pH 7.4 (Universal Buffer, Research Genetics, Huntsville, AL, U.S.A.), and digested with pepsin (2.5 mg/ml, Research Genetics) for 3 minutes at 105°C. Probe was applied in formamide-free diluent, and the slides were heated to 105°C for 3 minutes, cooled for approximately 1 minute at room temperature, and allowed to hybridize at 45°C for 60 minutes. The sections were then washed twice with 2× standard saline citrate at 45°C (3 minutes/wash) and detected with alkaline phosphatase conjugated streptavidin (Research Genetics). After washing once in AP Chromogen Buffer, pH 9.5 (Research Genetics) at room temperature, hybridization products were visualized with fast red. The slides were counterstained with hematoxylin, air dried, and coverslipped for microscopic examination.

Quantitative analysis of aromatase mRNA by PCR

The bone specimens were first crushed into fine pieces on dry ice. The methods have been previously described by the authors.12,28 Briefly, samples were carefully homogenized in 5 vol of 5 mol/l guanidine thiocyanate containing 5 mmol/l sodium citrate and 0.5% sodium sarcosyl at 4°C. The total RNA fraction from all homogenates was prepared as described by Chirgwin et al.29 The measurement of aromatase mRNA in these samples was performed by RT-PCR using a specific sense primer labeled with a fluorescent dye and a specific antisense primer as previously described.12,28,30 This quantitative analysis was a modification of the method originally reported by Wang et al.,31 which used a synthetic RNA as an internal standard and coamplified both the specific target mRNA and the internal standard in one reaction with the same primers. The internal standard RNA used in the assays was synthesized in vitro from modified aromatase cDNA with 0.01 attomol of human aromatase RNA containing a 21-base insertion and amplified by PCR for 26 cycles using a fluorescent dye (FAM)-labeled primer to examine the accuracy of this fluorescent quantitaion. A FAM-labeled sense primer (5′-TACTACAACCGGGTATATGG-3′, the sequence in exon 3) and an antisense primer (5′-TGTTAGAGGTGTCCAGCATG-3′, the sequence in exon 5) were used in the PCR for quantitative analysis of aromatase mRNA. An aliquot (4 μl) of the fluorescent PCR products was mixed with 3 μl of GENESCAN-1000 ROX consisting of DNA size markers labeled with a fluorescent dye, ROX, and analyzed fluorometrically with a Gene Scanner 362 (Applied Biosystems). The FAM-labeled PCR products showed two peaks corresponding to PCR products of aromatase mRNA and the internal standard RNA at positions of approximately 378 and 399 bp, respectively. These two peaks were designated AROM mRNA and Standard RNA, respectively (Fig. 1A). The fluorescent peak areas of these two PCR products were proportional to the amounts of both RNAs added as templates. This proportionality between their amounts and their fluorescent peak areas was detected over a wide range of 0.002–10 attomol in this method, as was reported previously.28 In addition, identification of the PCR products as aromatase mRNA was confirmed by its direct sequencing, and the fluorescent peak areas were also proportional to the quantity of tissue RNA added in the range of 1–20 μg.28 These findings indicate that optimal amounts of aromatase mRNA can be reasonably quantitated in tissue specimens by the fluorescent method. The Genescan-1000 ROX of the internal size standards gave seven peaks of 262, 293, 317, 439, 557, 691, and 695 bp, as is shown in Fig. 1A. The amount of aromatase mRNA in the tissue RNA was calculated from the peak areas of fluorescent products by the internal standard as previously described.12,28,30

Figure FIG. 1.

An example of analysis of RT-PCR analysis of aromatase mRNA content (A) and utilization of alternative exon 1 (B) by GeneScanner 362 in lumbar vertebra a 61-year-old female (case #16 in Table 1). Peaks A, B, C, D, E, F, and G represent the Gene Scan 1000 ROX of the internal size standards, 261, 293, 317, 439, 553, 691, and 695 base pairs, respectively. (A) Peaks I and II represents PCR products of aromatase mRNA and the internal standard RNA at positions of 378 and 399 base pairs, respectively. In this case, mRNA content was calculated as 284 attomol/ng of mRNA.

Utilization of alternative exons 1

The utilization of alternative exons 1 of the aromatase gene was examined by RT-PCR of the RNA fraction using sense primers specific for exons 1a, 1b, 1c, and 1d and the fluorescent dye-labeled antisense primer specific for exon 2 as described previously.12,28,30 Harada et al. have previously reported that exons 1a, 1b, 1c, and 1d are utilized in aromatase expression in human placenta, skin fibroblasts, and fetal liver, ovary, and prostate, respectively.17 Fluorescent PCR products were analyzed with a Gene Scanner 362. The aromatase mRNAs transcribed from exons 1a, 1b, 1c, and 1d yielded PCR products at positions corresponding to 402, 327, 368, and 355 bp, respectively. The Genescan-1000 ROX was used as internal size standards as in quantitative analysis of aromatase mRNA (Fig. 1B).


Immunolocalization of aromatase and type I 17β-HSD

Relatively weak aromatase immunoreactivity was observed in smooth muscle of the vascular wall and some osteocytes. Relatively strong aromatase immunoreactivity was observed in lining cells and osteoblasts (Fig. 2A), some chondrocytes of articular cartilage of tibia (Fig. 2B), and adipocytes adjacent to bone trabeculae (Fig. 2C). Among osteoblasts, aromatase immunointensity was much stronger in those with less cytoplasm. The majority of alkaline phosphatase positive osteoblasts (Fig. 2D) were also positive for aromatase. Patterns of type I 17β-HSD immunolocalization were also similar to that of aromatase (data not shown). Osteoclasts that demonstrated positive reaction in TRAP histochemistry were immunohistochemically negative for both aromatase (Fig. 2E) and type I 17β-HSD.

Figure FIG. 2.

Immunohistochemistry of aromatase (A,B,C) and alkaline phosphatase (D) and histochemistry of tartarate-resistant acid phosphatase (E) in the tibia of 60-year-old female. (A) Aromatase immunoreactivity was detected in osteoblast positive for alkaline phosphatase as shown in Fig. 2D (arrows). Weak signals were also detected in some ostocytes (×350). (B) Aromatase immunoreactivity was observed in chondrocytes in articular cartilage (×250). (C) Marked aromatase immunoreactivity was observed in lining cells or osteoblasts (arrows) and adipocytes (double arrows) but not in osteoclasts (arrow heads), positive for tartarate-resistant acid phosphatase as shown in Fig. 2E (×300). (D) Alkaline phosphatase immunoreactivity was shown by arrows (×300). (E) Tartarate-resistant acid phosphatase positive cells were shown by arrows (×300).

Aromatase mRNA expression by in situ hybridization

There was a marked heterogeneity in distribution of mRNA hybridization signals. Aromatase mRNA hybridization signals were identified primainly in lining cells (Fig. 3A), osteoblast chondrocytes of articular cartilage (Fig. 3B), and some adipocytes. Weak hybridization signals were also detected in some osteocytes (Fig. 3A). No significant accumulation of aromatase mRNA hybridization signals was observed in osteoclasts (Fig. 3C). Control sections hybridized with a sense oligonucleotide probe displayed no hybridization signals (Fig. 3D).

Figure FIG. 3.

In situ hybridization of aromatase mRNA in the tibia in 57-year-old female. (A) Aromatase mRNA hybridization signals appearing as a fast red reaction (arrows) were detected in lining cells and osteoblasts (×250). (B) Aromatase mRNA hybridization signals were also detected in chondrocytes in articular cartilage (×350). (C) Aromatase mRNA hybridization signals were detected in lining cells (arrows) but not in osteoclasts (double arrows) (×250). (D) Negative control with a sense oligonucleotide probe showed no detectable specific mRNA hybridization signals (×250).

Quantitation of aromatase mRNA by RT-PCR

Results are summarized in Table 1. Aromatase mRNA could be detected in 27/28 cases. Amounts of aromatase mRNA varied widely among the subjects (0.61–42.84 attomol/ng of total mRNA). The amount of aromatase mRNA of male subjects was 12.44 ± 8.25 attomol/ng of total RNA and that of female subjects was 9.67 ± 11.7 attomol/ng of total RNA. There was no significant difference between male and female subjects. There were no significant differences of the amount of aromatase mRNA among rib, vertebra, and femur. There was no correlation observed between the amount of aromatase mRNA and age of the subjects. The amount of aromatase expression in the subjects with a different degree of osteoporotic changes evaluated by radiological findings of vertebrae was as follows: normal (4), 7 ± 6.32 attomol/ng of total RNA; I (−) or osteopenia (10), 6 ± 4.49 attomol/ng total RNA; I (+) (7), 11.3 ± 7.99 attomol/ng of total RNA; II (7), 21.14 ± 11.9 attomol/ng of total RNA. A significant correlation was observed between the degree of osteoporotic changes and the amounts of aromatase mRNA (p < 0.05), i.e., the bone tissues obtained from the subjects with advanced osteoporotic changes had significantly higher aromatase mRNA expression than those not.

Alternative utilization of multiple copies of exon 1

Alternative utilization of multiple copies of exon 1 could be evaluated in 26/28 cases. Results are summarized in Table 1. The size of PCR products were 402 ± 3, 307 ± 3, 368 ± 3, and 355 ± 3 bp for exons 1a, 1b, 1c, and 1d, respectively, in this study. Exon 1b, the fibroblasts type exon, was the only promoter in 22 cases. (Table 1) Utilization of multiple exon 1 could be detected in four vertebrae and none in ribs and femurs examined. Among these four vertebrae, 1c, the gonadal type, was utilized as a major promoter in three cases.


Aromatase has been detected in many human extragonadal tissues7–9 including human bone.5,15 Estrogens derived from the local aromatization of testosterone or androgens have been considered to mediate many aspects of testosterone actions and exert estrogenic actions on these tissues. In this study, we quantitated aromatase mRNA by subjecting it to reverse transcription in the presence of an internal standard RNA and then amplifying the resulting cDNA by PCR with a fluorescent primer. Aromatase mRNA has also been quantitated by employing quantitating the RT-PCR products by titration of a cDNA template against a series of dilutions of a standard template.32,33 We used the fluorescent method because the method is simple and can avoid radioactivity. However, direct comparison of these two different methods in the same specimens, i.e., competitive PCR amplification and this fluorescent method, is required for clarification.

In human bone, Schweikert et al. recently studied aromatase activity by the tritiated water method in femur of 11 women and 4 men and demonstrated that the bone from both men and women had the capacity to form estrogen from androstenedione, although the values varied widely among the subjects (0.14–1.23 nmol/g DNA/h).15 This result is consistent with our findings, i.e., the presence of aromatase mRNA in almost all the subjects of different age groups examined, marked variation of the amount of aromatase mRNA among the subjects and no significant differences between male and female subjects. However, the possibility that the wide variation of the amount of aromatase mRNA may be due to the quantitative method employed cannot be completely ruled out. Results of our present study and Schweikert et al.15 indicated that human bone tissues can produce estrogen in situ through aromatization. The aromatization of androstenedione was demonstrated in cultured normal human osteoblast-like cells and osteoblast-like osteosarcoma cells.5,23,26 The presence of aromatase mRNA was also reported in these cultured cells.5 The presence of aromatase immunoreactivity and mRNA hybridization signals in lining cells and osteoblasts is consistent with these findings above.5,23,26 In addition, type I 17β-HSD was also expressed in these cells, which indicates that estrone, produced as a result of aromatization, is actively converted to estradiol in lining cells and osteoblasts. Human osteoblast-like cells were also reported to contain estrogen receptor.34,35 Therefore, locally produced estradiol may act directly at their site of synthesis. Aromatase expression was also observed in chondrocytes in articular cartilage, suggestive of a possible involvement of in situ estrogen production in chondrocytes cells.

Nawata et al. recently demonstrated a positive correlation between serum dehydroepiandrostenedione levels and bone mineral density in postmenopausal women.5 However, the great majority of previous studies failed to demonstrate the correlation between the total levels of circulating androgens and the degree of osteoporosis in postmenopausal women.15,36,37 Therefore, it is very interesting to examine whether there are any correlations between the amount of aromatase and the degree of osteoporotic changes. We studied aromatase mRNA expression using RT-PCR in the bone specimens retrieved from tissue banks stored at −80°C. Therefore, the specimens for determining bone mineral density or evaluating morphometrical analysis were not available for our study. Therefore, although not as accurate as those methods above, we attempted to evaluate retrospectively the presence or absence and/or the degree of osteoporotic changes in radiological findings of vertebrae of the patients retrieved from the charts. Therefore, there are some limitations in interpreting the data but bone tissues from the patients with more advanced osteoporotic changes had significantly higher amounts of aromatase mRNA expression, altough there was no correlation between age of the patients and the amount of aromatase mRNA. However, it is also true that marked aromatase expression was also detected in adipocytes, which increase in the number in osteoporotic bone. Therefore, the positive correlation between radiological osteoporotic changes and the amount of aromatase mRNA may represent an increase of adipocytes in these bones but further investigation are required.

Many reports have indicated that the expression of aromatase in various human tissues is regulated tissue-specifically by different factors, chiefly alternative use of exon 1 as promoters.16–18 Results of our present study demonstrated that human bone tissues predominantly utilized exon 1b (fibroblast type), and utilization of exon 1c (gonadal type) and multiple copies of exon 1 were detected in some vertebrae bone. Recently, switching of alternative exons 1 was proposed to result in increased aromatase expression and possible overproduction of estrogen in situ under the control of a new promoter.38 However, results of our study indicated that this switching does not play a major role in aromatase overexpression in human bone tissue. Because this is a small and rather limited study, further investigations are required to clarify the mechanism by which estrogens formed in situ influence human bone metabolism especially with relation to the development of osteoporosis.


This work was supported in part by a grant from the Ministry of Education, Japan and by a grant from Public Trust Haraguchi Memorial Cancer Research Fund, Tokyo, Japan.