Estrogen Stimulates Degradation of β-Amyloid Peptide by Up-regulating Neprilysin*

Postmenopausal estrogen depletion is a characterized risk factor for Alzheimer disease (AD), a human disorder linked to high levels of β-amyloid peptide (Aβ) in brain tissue. Previous studies suggest that estrogen negatively regulates the level of Aβ in the brain, but the molecular mechanism is unknown. Here, we provide evidence that estrogen promotes Aβ degradation mainly through a principal Aβ degrading enzyme, neprilysin, in neuroblastoma SH-SY5Y cells. We also demonstrate that up-regulation of neprilysin by estrogen is dependent on both estrogen receptor α and β (ERα and ERβ), and ligand-activated ER regulates expression of neprilysin through physical interactions between ER and estrogen response elements (EREs) identified in the neprilysin gene. These results were confirmed by in vitro gel shift and in vivo chromatin immunoprecipitation analyses, which demonstrate specific binding of ERα and ERβ to two putative EREs in the neprilysin gene. The EREs also enhance ERα- and ERβ-dependent reporter gene expression in a yeast model system. Therefore, the study described here provides a putative mechanism by which estrogen positively regulates expression of neprilysin to promote degradation of Aβ, reducing risk for AD. These results may lead to novel approaches to prevent or treat AD.

Alzheimer disease (AD) 3 is a progressive neurodegenerative disease characterized by declarative memory impairment and progressive dementia. The level of ␤-amyloid peptide (A␤) is elevated in the brains of AD patients, and A␤ is believed to play a critical role in the pathology of AD (1,2). Recent studies show that aggregated oligomers of A␤ (protofibrils) play a direct role in neuronal and behavioral deficits in AD patients (3).
The rate of A␤ degradation could influence the risk of developing AD, and it has been proposed that stimulation of proteolytic degradation of A␤ could be used as a therapeutic approach for AD (4,5). Neprilysin is thought to be the primary A␤-de-grading enzyme in the brain (6) because degradation of radiolabeled synthetic A␤42 in rat brain is largely inhibited by the neprilysin inhibitor, phosphoramidon (PA) (7,8) and because neprilysin degrades both monomeric and oligomeric forms of A␤40 and A␤42 in intracellular and extracellular compartments of the brain (9). Moreover, the level of neprilysin mRNA and protein is lower in the hippocampus and temporal gyrus of AD patients (10,11), which correlates with higher levels of A␤ as A␤ tends to accumulate in these regions (12).
Neprilysin activity is also lower in the hippocampus, cerebellum, and caudate of ovariectomized rats than in non-ovariectomized rats, and this effect can be reversed by exogenous 17␤estradiol (13). These data indicate that 17␤-estradiol positively regulates neprilysin activity in the brain. 17␤-Estradiol was also reported to reduce the generation of A␤ peptides in neuroblastoma cells and neurons (14). The positive regulation of neprilysin by 17␤-estradiol might be a crucial factor in protecting normal adult brain from A␤ damage and in improving cognitive performance in menopausal women.
This study reports that 17␤-estradiol promotes A␤ clearance by up-regulating neprilysin expression in human neuroblastoma SH-SY5Y cells. In these cells, 17␤-estradiol stimulates neprilysin expression in an estrogen receptor (ER)-dependent manner. Furthermore, two functional estrogen response elements (EREs) were identified in the neprilysin gene, which bind ER␣ and ER␤ in vitro and in vivo and which stimulate ER-dependent reporter gene expression in a yeast system. These results provide insight into the neuroprotective effects of estrogen and suggest that neprilysin could have potential as a therapeutic drug target for AD.
Recombinant Plasmids-YEP-ER␣ and YEP-ER␤, which express ER␣ and ER␤ in yeast, respectively, YRPC2, containing a CYC1 promoter and a lacZ reporter gene, and YRPC2-2cERE, which has two consensus EREs (cEREs) in front of the CYC1 promoter, were gift from Dr. Dan Noonan (University of Kentucky). Three putative EREs were identified in the neprilysin genomic sequence. Oligonucleotides containing two copies of each putative ERE and a terminal XhoI recognition site were synthesized (Table 1), annealed, phosphorylated, and ligated into YRPC2 plasmid. Recombinant plasmids pET-28b-ER␣ and pET-28b-ER␤ that expressed ERs in Escherichia coli Rosetta (DE3) cells were constructed by inserting ER-coding sequences from YEP-ER␣ and YEP-ER␤ at the NdeI and EcoRI sites. pEGFP-ER␣ and pEGFP-ER␤ that expressed ERs in mammalian cells were obtained by inserting ERs-coding sequences at the XhoI and EcoRI sites.
A␤42 Degradation Assay and Neprilysin Activity Assay-SH-SY5Y cells were treated with 100 nM 17␤-estradiol or vehicle. 24 h later the medium was removed, and the cells were incubated with medium spiked with A␤42 (100 pg/ml; Sigma) and 1 M ZnCl 2 without the serum supplemented. The medium was collected 8 h later, and the A␤42 levels were detected by enzymelinked immunosorbent assay kits (CUSABIO BIOTECH). After 17␤-estradiol treatment for 24 h, SH-SY5Y cells were harvested, and the membrane proteins were extracted by protein extract kit (DBI Research Products). 150-g membrane fractions were evaluated for neprilysin enzymatic activity as described previously by Huang et al. (13). PA (Sigma) was used as a specific neprilysin inhibitor to block the neprilysin activities.
RNA Isolation and RT-PCR-Total RNA was extracted using TRIZOL reagent from Invitrogen. cDNA was synthesized using a ReverTra Ace-␣-cDNA synthesis kit (TOYOBO) and random primers (TaKaRa) following the manufacturer's instructions. Semiquantitative PCR was performed to determine the mRNA levels of neprilysin. Assessment of the PCR conditions ensures that the cycle is in the linear range.
Small Interfering RNA (siRNA) Duplexes Specific for Human ER␣ and ER␤-Two siRNA duplexes specific for human ER␣ and ER␤ were designed according to Musatov et al. (15). The targeted 19-mer oligonucleotides sequence for ER␣ and ER␤ were GGCATGGAGCATCTCTACA (1923-1941 nucleotides, NM_000125.2) and GGCATGGAACATCTGCTCA (1882-1900 nucleotides, NM_001437.2). All siRNA sequences were designed and synthesized according to the manufacturer's recommendations (GenePharma). The siRNA duplex (UUCUCC-GAACGUGUCACGUTT and ACGUGACACGUUCGGAG-AATT), which had no significant homology to any known gene sequences from mouse, rat, and human, served as a negative control.
RNA Interference and Overexpression of ERs-The transfection of siRNA duplexes into SH-SY5Y cells cultured in six-well plates was performed using Lipofectamine 2000 (Invitrogen) as described by Deng (16). 100 nM siRNA duplexes were transfected to knock down ER␣ or ER␤, and 48 h later 10 nM 17␤estradiol was added to induce the expression of neprilysin.
Delivery of 1 g of recombinant pEGFP plasmids to SH-SY5Y cells grown in 6-well plates was carried out using Lipofectamine 2000 (Invitrogen) following the manufacturer's instructions. 48 h later 10 nM 17␤-estradiol was supplemented to induce the expression of neprilysin.
Transcription Factor Binding Site Analysis-The neprilysin gene sequence is available from the NCBI nucleotide data base (www.ncbi.nlm.nih.gov) (GenBank TM accession number AC106724, Homo sapiens 3 BAC RP11-270G15 complete sequence) (Roswell Park Cancer Institute Human BAC Library). The putative EREs were determined using TRANSFAC 6.0.
Expression and Purification of Recombinant ERs-pET-28b-ER␣/ER␤ was transformed into E. coli Rosetta (DE3) cells. Transformed cells were cultured in Luria-Bertani medium supplemented with kanamycin (40 g/ml) and chloramphenicol (34 g/ml) at 37°C until A 600 reached 0.8. Then 0.5 mM isopropyl 1-thio-␤-D-galactopyranoside was added to induce the expression of recombinant His 6 -ERs for 24 h at 20°C. Recombinant His 6 -ERs were purified on nickel-Sepharose (GE Healthcare) according to the GE Healthcare protocol. The column was washed with column balance buffer containing 40 mM imidazole, and the bound protein was eluted with 260 mM imidazole. Dialysis was applied to remove imidazole from the eluants.

TABLE 1 Sequences of oligonucleotides for construction of YRPC2 plasmids containing putative EREs
Initial, single-underlined (TCGA) sequences indicate XhoI sites. Double-underlined sequences are sequences of putative EREs. Bold residues are core sequences of putative EREs.

Name
Direction Seqence (5-3) Nonradioactive Gel Mobility Shift Assay-Digoxygenin-labeled probes containing putative EREs or cERE was prepared by amplification of recombinant YRPC2 plasmids with the primers 5Ј-digoxigenin-TGATCATGTGTCGTCGCA-3Ј and 5Ј-TCTGAGTTCCGACAACAATG-3Ј. Labeled probes were PAGE-purified. Nonradioactive gel mobility shift assay was performed according to Hammer's Protocol for Nonradioactive electrophoretic mobility shift assays. 50 ng of end-labeled oligonucleotide and 50 g of purified His 6 -ER were mixed in the binding reaction and incubated for 60 min. After analyzed on 6% non-denaturing polyacrylamide gels and transferred to nylon membrane, the probes were blotted with the anti-digoxigenin-alkaline phosphatase Fab fragment (Roche Applied Science). Ready-to-use substrate nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate solution (Amresco) was used to develop the nylon membrane.
Yeast Transformation and ␤-Galactosidase Assay-Recombinant YRPC2 plasmids containing the putative EREs and YEP-ER␣ or YEP-ER␤ were transformed into BJ5409 yeast cells using quick and easy TRAFO Protocol (17). ␤-Galactosidase assays were performed essentially as we described before (18).
Statistical Analysis-Results are expressed as the mean Ϯ S.D. Statistical analysis was performed using one-way analysis of variance followed by the Bonferroni ad hoc test. The intensities of the bands were quantitated using Quantity One software (Bio-Rad).

17␤-Estradiol Enhances A␤42 Degradation and Up-regulates
Neprilysin-Previously we demonstrated that 17␤-estradiol stimulates expression of neprilysin in the hippocampus, cerebellum, and caudate of rat brain (13). Here, we measured degradation of exogenous A␤ in SH-SY5Y cells, a human neuroblastoma cell line widely used for A␤ metabolism (19). As expected, exogenous A␤42 was degraded more efficiently (reduced to 36% of input substrate) in cells treated with 17␤estradiol (100 nM for 24 h). Increased degradation of A␤42 was blocked by PA, a specific neprilysin inhibitor (reversed to 73%) (Fig. 1A). In parallel with enhanced degradation of A␤, neprilysin activity increased ϳ8.5-fold, and this increase was 80% inhibited by PA (Fig. 1B). These results suggest that 17␤-estradiol might up-regulate expression of neprilysin, which in turn stimulates degradation of A␤. . SH-SY5Y cells were grown in 17␤-estradiol-free medium at least 3 days before use. A, after incubation with 100 nM 17␤-estradiol or vehicle for 24 h, cells were incubated with A␤42 (100 pg/ml). A␤42 degradation and the reverse effect by PA (75 M) were detected by enzyme-linked immunosorbent assay kits. Medium without A␤42 was used as a negative control. B, cells were treated with vehicle or 100 nM 17␤-estradiol for 24 h, harvested, and lysed. 150-g membrane fractions were evaluated for neprilysin enzymatic activity. PA (75 M) was used to specifically inhibit neprilysin activity. C, cells were treated with vehicle, 100 pM, 1 nM, or 10 nM 17␤-estradiol for 6 h. Total RNA was prepared, and RT-PCR was carried out as described under "Materials and Methods." Glyceraldehyde-3-phosphate dehydrogenase (G3PDH) was an internal control. D, cells were incubated with vehicle, 100 pM, 1 nM, or 10 nM 17␤-estradiol for 24 h, harvested, and lysed. Total cellular protein was analyzed by Western blot using anti-CD10 and anti-␤-actin (internal control). E, cells were treated with 10 nM 17␤-estradiol for 0, 1, 6, 12, or 24 h. Total RNA was prepared, and RT-PCR was carried out. F, cells were treated with 10 nM 17␤-estradiol for 0, 1, 6, or 24 h. Western blots were probed as indicated. Data values are represented as the mean Ϯ S.D., and three independent experiments were performed. **, p Ͻ 0.01.
RT-PCR and Western blot analysis showed that 17␤-estradiol increases neprilysin mRNA and protein in a dose-and timedependent manner. Neprilysin mRNA increased Ͼ2-fold in cells treated for 6 h with 1 nM 17␤-estradiol and Ͼ5-fold in cells treated with 10 nM 17␤-estradiol (Fig. 1C). The level of neprilysin protein increased significantly in cells treated for 24 h with 1 or 10 nM 17␤-estradiol (p Ͻ 0.01) (Fig. 1D). Neprilysin mRNA reached and was maintained at a maximal level ϳ6 h after treatment with 10 nM 17␤-estradiol (Fig. 1E), whereas neprilysin protein reached a maximum ϳ24 h after treatment (Fig. 1F). These results are consistent with the estrogen activation model proposed by Shang (20), namely, that neprilysin mRNA is relatively unstable, such that ongoing transcription is required to maintain the steady state level of neprilysin mRNA and protein.
ER␣/ER␤-selective Agonists Induce Neprilysin Expression-SH-SY5Y cells express both ER␣ and ER␤ (21). Therefore, the roles of ER␣ and ER␤ in regulating of estrogen-stimulated neprilysin expression were examined. This was done by incubating cells in the presence or absence of ER␣-specific agonist, PPT (22), or ER␤-specific agonist, DPN (23). As shown in Fig. 2, both PPT and DPN induced neprilysin mRNA ( Fig. 2A) and protein (Fig.  2B) in SH-SY5Y cells, and neprilysin activity was elevated by PPT and DPN (Fig. 2C). These subtype-specific agonists were less potent than 17␤-estradiol in inducing neprilysin, which is consistent with the fact that 17␤-estradiol activates both receptor subtypes. Interesting, 17␤estradiol also appears to stimulate expression of ER␣ and ER␤ in SH-SY5Y cells (Fig. 2D). These results suggest that both ER␣ and ER␤ mediate the effects of 17␤-estradiol on expression of neprilysin.
Induction of Neprilysin Is Dependent on ER␣ and ER␤-To further investigate the effects of ER subtypes in this process, the expression of ER␣ or ER␤ was up-or down-regulated by transient transfection with ER expression vectors or siRNA-mediated knockdown in SH-SY5Y cells. To overexpress ERs, GFP-tagged ER␣ or ER␤ was expressed from appropriate plasmid vectors, and ER expression was monitored by confocal microscopy (data not shown). As shown in Fig. 3, A and B, transfection of the GFP vector had little effect on neprilysin induction, whereas transfection of GFP-ER␣ or GFP-ER␤ induced neprilysin at least 3-fold.
Previously characterized ER␣or ER␤-targeted siRNAs (15) were used to knock down endogenous ER␣ or ER␤ by Ն70% in

. Overexpression and silence of ERs altered the 17␤-estradiol inducibility of neprilysin.
A, SH-SY5Y cells were transiently transfected with 1 g of pEGFP, pEGFP-ER␣, or pEGFP-ER␤ and incubated for 48 h. The inducibility of neprilysin was detected by anti-CD10 after 10 nM 17␤-estradiol (E2) administration for 24 h. The blots were quantified by densitometry (B). C, silence of ER␣ or ER␤ was conducted by transfection with 100 nM corresponding siRNA duplexes. The relative ERs levels were detected by anti-ERs. NC, negative control as described under "Materials and Methods." D, after 10 nM 17␤-estradiol administration for 24 h, the inducibility of neprilysin was detected by anti-CD10 and quantified by densitometry (E). Three independent experiments were performed. The ratios (neprilysin:␤-actin) were calculated and normalized to 17␤-estradioltreated (B)/untreated (E) and non-transfection groups. **, p Ͻ 0.01. SH-SY5Y cells (Fig. 3C). As shown in Fig. 3, D and E, single siRNA knockdown of ER␣ or ER␤ decreased the level of neprilysin slightly (19% or 44%), whereas a much larger effect (86% decrease) was observed when both ER␣ and ER␤ were knocked down. These results confirm that both ER␣ and ER␤ mediate the effects of 17␤-estradiol on expression of neprilysin.

Identification of Putative EREs in the Neprilysin Gene and Their Interaction with Recombinant ER␣ and ER␤ in
Vitro-One mechanism to explain the above data is that ligandactivated ER binds to target EREs in the neprilysin gene, which stimulates transcription of neprilysin mRNA. Typically, ligand-activated ER dimerizes, binds to target EREs, recruits RNA polymerase II, and stimulates transcription. This possibility was tested by computationally searching for homology to the cERE, which has the sequence GGT-CAnnnTGACC, in the neprilysin gene. This transcription factor binding site analysis identified three putative EREs (ERE-I, ERE-II, and ERE-III) with Ͼ80% similarity to the cERE (Fig. 4A). ERE-I and ERE-II are located upstream of the promoter, and ERE-III is in the ninth intron.
To determine whether ER␣ and ER␤ bound to putative EREs I-III from the neprilysin gene in vitro, gel mobility shift assays were performed using recombinant His 6 -ER␣ or His 6 -ER␤, which was expressed in and purified from Rossetta (DE3) E. coli (Fig. 4B). DNA substrates containing cERE, ERE-I, ERE-II, or ERE-III were generated by PCR amplification of target regions from the appropriate plasmid template. As expected, ER␣ and ER␤ did not bind to the negative control DNA probe but did bind to and alter the mobility of a DNA substrate containing cERE (Fig. 4C). Furthermore, ER␣ and ER␤ bind specifically to putative ERE-I, ERE-II, and ERE-III, altering the mobility of a large fraction of the input DNA substrate (Fig. 4D).
In Vivo Occupancy of EREs in the Neprilysin Gene in SH-SY5Y Cells-The significance of the putative EREs in regulating neprilysin expression can only be assessed by testing their functional effects in vivo. Therefore, ChIP assays were performed to evaluate occupancy of the three putative EREs in SH-SY5Y cells. In this experiment SH-SY5Y cells were treated for 45 min with vehicle or 17␤-estradiol, and cell lysates were immunoprecipitated using anti-ER antibody or control IgG. Neprilysin gene regions containing ERE-I, ERE-II, or ERE-III were amplified by semiquantitative PCR, as shown in Fig. 5A. ChIP results suggest that ER␣ and ER␤ occupy ERE-I and ERE-III in a 17␤estradiol-dependent manner in SH-SY5Y cells, but ER␣ and ER␤ do not bind ERE-II (Fig. 5B). Thus, there is a discrepancy between in vitro (i.e. gel mobility shift) and in vivo (ChIP assay) binding data for ERE-II, indicating that ERE-II may interact more weakly with ER␣ and ER␤ than ERE-I and ERE-III or that ERE-II may not play a direct role in promoting estrogen-dependent neprilysin expression in vivo. ChIP analysis also showed that the Ϫ121 to ϩ28 region of the neprilysin promoter is not bound by ER␣ or ER␤ in 17␤-estradiol-treated SH-SY5Y The number under or above each putative ERE is its distance from the transcription start site. The detailed sequences of EREs were also listed below. B, recombinant His 6 -ERs were expressed in Rossetta (DE3) E. coli cells, purified by Ni-Sepharose affinity chromatography, and characterized by immunoblotting with anti-ER␣ and anti-ER␤. C, the probe from the YRPC2 vector was used as negative control (PC2), and the probe from YRPC2-cERE was used as positive control (ERE). Recombinant ER␣ or ER␤ was included as indicated. D, putative EREs containing probes were incubated with ER␣ or ER␤ as indicated.

FIGURE 5. Occupancy of target sites in neprilysin gene in SH-SY5Y cells by ER␣ and ER␤.
The cells were grown in estrogen-depleted medium for at least 3 days before use. Cells were treated with 10 nM 17␤-estradiol (E2) or vehicle for 45 min, fixed, and harvested. ChIP assay was performed using rabbit anti human IgG, anti-ER␣ antibody, and anti-ER␤ antibody as indicated. Input was used as positive control. A, a map of neprilysin gene and locations of regions amplified for ChIP analysis is shown. ϩ1 is the transcription start site, and the numbers indicate the coordinates of amplified regions in the neprilysin genomic sequence. B, ChIP analysis of putative EREs and promoter region are shown. At least three independent experiments were performed; a representative result is shown. cells (Fig. 5B). These data support the idea that ERE-I and ERE-III, but not ERE-II, are functional EREs that promote 17␤-estradiol-stimulated ER-dependent induction of neprilysin in vivo.
Putative EREs Mediate ER␣and ER␤-dependent Reporter Gene Activity in Yeast-The function of EREs I, II, and III from the neprilysin gene was tested using a yeast reporter gene system. The putative EREs were cloned into the reporter plasmid YRPC2 to generate three recombinant plasmids: YRPC2-ERE-I, YRPC2-ERE-II, and YRPC2-ERE-III. YRPC2-cERE, which carries two copies of cERE, and YRPC2 vector were used as positive and negative controls, respectively. Recombinant YRPC2 plasmids and ER␣ or ER␤ expression plasmids (YEP-ER␣ and YEP-ER␤) were transformed into yeast BJ5409 cells. Fig. 6A shows the mechanism of the transcriptional activity assay. The ␤-galactosidase activities indicate the reporter gene activity. Because ER expression plasmids are exogenously expressed in yeast cells, this reporter gene system is free of interference from endogenous ER coregulators. Control experiments showed no reporter gene activity for any of the above constructs in 17␤estradiol-treated cells that do not express ER␣ or ER␤ (Fig. 6B). This result demonstrates the absolute dependence of reporter gene induction on exogenous ER.
However, in 17␤-estradiol-treated cells that express ER␣, all of the putative EREs showed reporter gene activity. ERE-I, ERE-II, or ERE-III constructs induced 9.6-, 3.7-, or 5.7-fold higher reporter gene activity than the same cells without 17␤-estradiol treatment, respectively (Fig. 6C). In 17␤-estradiol-treated cells that express ER␤ (Fig. 6D), ERE-I, ERE-II, or ERE-III constructs supported a 4.2-, 3.7-, or 6.3-fold increase in reporter gene activity. Thus, these data suggest that ligand-activated ER␣ and ER␤ independently activate reporter gene transcription with promoters containing ERE-I, ERE-II, or ERE-III. ERE-II appears to be a weaker enhancer than ERE-I and ERE-III. There was no evidence for synergism between ER␣ and ER␤ during activation of the three putative EREs (data not shown).

DISCUSSION
This study demonstrates for the first time that 17␤-estradiol and two selective ER␣/ER␤ agonists (PPT and DPN) positively regulate expression of neprilysin, which in turn stimulates A␤42 degradation in human neuroblastoma SH-SY5Y cells. The data presented here also show that 17␤-estradiol stimulates expression of neprilysin in an ER␣and ER␤-dependent manner. Furthermore, we identify two novel putative EREs (ERE-I and ERE-III) located in the neprilysin genomic region that bind ER␣ and ER␤ in vitro and in vivo and convey 17␤-estradiol inducibility of a reporter gene in yeast. Together these results indicate a possible novel mechanism for the neuroprotective activity of estrogen.
Previous studies show that estrogen blocks the neurotoxic effects of A␤42 in SH-SY5Y cells (21). These observations are consistent with our report in this study that 17␤-estradiol decreases the neurotoxic A␤42 in SH-SY5Y cells. Furthermore, we observe that the decrease can be reversed by the specific neprilysin inhibitor (Fig. 1A). Because neprilysin is the principal A␤ degrading enzyme, even modest up-regulation of neprilysin can reduce accumulation of A␤ in the mouse brain (6,24), or even partial down-regulation of neprilysin activity can contribute to AD development by promoting A␤ accumulation (7). This is consistent with our observation that after 17␤-estradiol treatment, the inhibition of neprilysin by the specific neprilysin inhibitor correlates with decreased A␤ degradation. Thus, 17␤estradiol may prevent A␤-associated neurotoxicity through its ability to stimulate neprilysin-mediated degradation of A␤ peptides. These results provide a possible explanation for the observation that women who receive estrogen replacement therapy soon after menopause have a decreased risk for AD as well as higher cognitive performance.
ER␣ and ER␤ have significant neuroprotective effects in neuronal cells (25,26). PvuII and XbaI polymorphisms located in ER␣ (27)(28)(29) and five intronic single-nucleotide polymorphisms (30) in the ER␤ gene were identified as susceptibility FIGURE 6. Putative EREs could induce yeast reporter gene expression with ER␣ and ER␤. BJ 5409 yeast cells were transformed with reporter gene constructs as indicated, grown to A 600 ϭ 0.60, and treated with 100 nM 17␤-estradiol (E2) for 24 h. ␤-Galactosidase activity was an indicator of reporter gene activity. A, shown is a map of recombinant reporter gene plasmids and the basic mechanism of this transcriptional activity assay. B, shown is ␤-galactosidase activity in yeast cells that do not express ERs. C, shown is ␤-galactosidase activity in yeast cells expressing ER␣. D, shown is ␤-galactosidase activity in yeast cells expressing ER␤. Three independent experiments were performed. **, p Ͻ 0.01. factors for AD in women. Most interestingly, in the hippocampus of female rats, expression of ERs decreases with age (31,32). If a similar decrease occurs in human brain, it might explain the decreased efficacy of estrogen replacement therapy in older postmenopausal women (33). In this study we show that selective ER agonists (PPT and DPN) positively and independently regulate neprilysin expression (Fig. 2) and that overexpression of either ER␣ or ER␤ significantly enhances the 17␤-estradiol inducibility of neprilysin. Also, siRNA knockdown of ER␣ and ER␤ significantly inhibits induction of neprilysin (Fig. 3). Therefore, 17␤-estradiol stimulates neprilysin gene expression in an ER-dependent manner, and each ER subtype independently mediates such stimulation.
Previous studies demonstrated that 17␤-estradiol exerted its transcriptional effects through genomic pathways and/or nongenomic pathways as reviewed by Björnström and Sjöberg (34). In this study we have identified two putative EREs in the neprilysin gene, ERE-I and ERE-III, that can bind ERs in vitro (Fig. 4) and can be occupied by ER␣ and ER␤ in vivo (Fig. 5). These two EREs confer 17␤-estradiol inducibility in a yeast reporter gene system (Fig. 6), suggesting that 17␤-estradiol stimulates ERs, which bind EREs in the neprilysin gene and recruit coactivators to initiate neprilysin expression.
Deschênes et al. (35) found that chromatin loops form between multiple EREs spanning 20 kb and a transcriptional start site in the presence of 17␤-estradiol, suggesting that a multipartite transcription-regulatory complex could play a role in regulating ER target genes. In our study, although ERE-I is far upstream of the neprilysin promoter and ERE-III is in the ninth intron, it is possible that chromatin loops form between the two EREs, and the neprilysin transcriptional start site in the presence of 17␤-estradiol and a similar multipartite transcriptionregulatory complex play a role in initiating neprilysin transcription (20,36).
Last year, Yao et al. (37) found that androgen up-regulates neprilysin gene expression, and their study suggests that androgen treatment could reduce accumulation of A␤, as a therapeutic approach for AD. This study demonstrates positive regulation of neprilysin expression by 17␤-estradiol, leading to A␤ degradation. Undoubtedly, AD may be correlated with the declining level of sex hormones; in fact, depletion of 17␤-estradiol has long been regarded as a crucial risk factor for AD in postmenopausal woman. Hence, our data about the regulation mechanism of A␤ and neprilysin by 17␤-estradiol may provide a clue to understanding more about the onset of AD and its relation with sex hormones.
However, therapeutic use of estrogen in AD patients could also have adverse effects (such as endometrial carcinoma and breast cancer). The multitarget effect of estrogen may be influenced by distinct expression and differential activities of ER␣ and ER␤ (38). Interestingly, adverse effects of exogenous estrogen tend to involve ER␣ (39). On the other hand, ER␤ activation correlates with stronger anti-tumor activity (40). Some ER␤selective chemicals (such as genistein) have fewer adverse effects than 17␤-estradiol, but confer comparable therapeutic benefits. Most importantly, in cerebral cortex and hippocampus, which are directly related to cognitive behavior and play a greater role in AD pathology (41), ER␤ is the predominant expressed ER (42,43). Results presented here suggest that ER␤ independently mediates 17␤-estradiol stimulated neprilysin expression (Figs. 2 and 3). These results suggest that ER␤-selective compounds may be extremely valuable agents for A␤ degradation with less adverse effects. Because depletion of 17␤estradiol is an established risk factor for AD in post-menopausal women, the results of this study may have important clinical implications in prevention and/or treatment of AD.