Sex-specific Expression Mechanism of Hepatic Estrogen inactivating Enzyme and Transporters in Diabetic Women

Abstract

Circulating estrogens levels significantly decrease in menopause and levels off in postmenopausal women. Accordingly, the liver represses levels of enzymes and membrane transporters, thereby decreasing capability of inactivating and excreting estrogens. Women increasingly develop type 2 diabetes during or after menopause. Estrogens are known to promote liver diseases in these women. Here, we have found that the estrogen inactivating sulfotransferase (SULT1E1) and an efflux transporter ATP-binding cassette subfamily G member 2 (ABCG2) that exports sulfated estrogens increased their expression levels in diabetic women but not men. For the sulfotransferase gene, phosphorylated nuclear receptors ERα and RORα, at Ser212 and Ser100, respectively, bind their response elements to activate the SULT1E1 promoter in women. This coordinated increase in estrogen inactivation and excretion, and the phosphorylated nuclear receptor-mediated gene activation could be a defense mechanism against toxicities of estrogens through inactivation and excretion in the livers of women.

1. Introduction

Type 2 diabetes (T2D) is a metabolic disease characterized by elevated blood sugar due to impaired insulin response [1]. Diabetes complicates many organs such as the liver where the disease promotes inflammation, fibrosis, and cirrhosis [2, 3]. Women become prone to develop T2D during menopause or in post menopause in part due to declining estrogens. Although estrogens enhance efficient energy metabolism [4–8] and demonstrate anti-inflammatory effects [9, 10], the hormone can promote complications in postmenopausal women [11–13]. Therefore, proper regulation of estrogen levels can be critical to suppress the complications. Estrogens are metabolically inactivated by estrogen sulfotransferase (SULT1E1) that effectively sulfates estrogens [14, 15], and sulfated estrogens are stored in circulating blood or exported into the bile duct for elimination [16–18]. Hepatic membrane transporters organic anion transporting polypeptide 1B1 (OATP1B1, encoded by solute carrier organic anion transporter family member 1B1/SLCO1B1) and breast cancer resistance protein (BCRP, encoded by ATP-binding cassette superfamily G member 2/ABCG2 gene) mediate these import and export of estrogens and their derivatives [19–22], respectively.

The liver is the major organ for estrogen metabolism and disposition [14, 15, 23]. Age-, sex- and diabetes-dependent expressions of the SULT1E1 in human livers have not yet been comprehensively studied. However, studies with mice could be indicative of how the SULT1E1 is expressed in humans. In mice, Sult1e1 is expressed after birth but its expression levels are maintained at low levels in both adult males and females [24, 25]. On the other hand, the Sult1e1 expression is elevated in both males and females of diabetic mice [25]. Therefore, both basal and diabetes response expressions are, in principle, not sex dependent. In addition to diabetes, phenobarbital (PB), a sedative used for epilepsy, and garlic extract diallyl sulfide also induced hepatic expression in mouse livers [26, 27].

The molecular mechanism of the activation Sult1e1 gene has been determined in PB treated or diabetic mouse livers. This induction is mediated by nuclear receptors Constitutive androstane receptor (CAR, NR1I3), estrogen receptor alpha (ERα, NR3A1) and retinoic acid receptor-related orphan receptor alpha (RORα,NR1F1) in the diabetic mouse liver [25, 26]. These nuclear receptors utilize conserved phosphorylation sites within their DNA binding domains (DBD) to integrate their functions, activating the Sult1e1 gene [25, 26]. Apparently, the phosphorylation-mediated mechanism which functionally integrates these nuclear receptors is conserved to activate the Sult1e1 in mouse livers. In humans, the SULT1E1 expression decreases in liver diseases such as diabetic cirrhosis and in alcoholic cirrhosis tissues [28]. Whether the SULT1E1 expression is sex-dependent in diabetic human liver remains unknown although its expression is sex-independent in mice. Moreover, expressions of influx and efflux membrane transporters associated with estrogen import/export have not been studied concerning estrogen metabolism and disposition in diabetic human livers. Determining these expressions, and mechanisms should be insightful in understanding estrogen metabolism and disposition, and biology in diabetic human livers.

Here, we examined the expressions of genes encoding the SULT1E1 and membranes transporters SLCO1B1 and ABCG2 in diabetic and nondiabetic human liver biopsies and human primary hepatocytes (HPH). Moreover, the nuclear receptor-mediated mechanism of a diabetes induced activation the SULT1E1 gene was also examined. Comparative RT-qPCR analysis of SULT1E1, ABCG2, and SLCO1B1 mRNA expression levels was performed between nondiabetic and diabetic men and women. To examine the nuclear receptor-mediated regulatory mechanisms, CAR response element (DR4), RORα response element (RORE), and ERα response element (ERE) were determined in a distal region of the SULT1E1 promoter. Moreover, using cell-based reporter assays, a proximal element (PE) on the SULT1E1 promoter appeared to be vital for the transcriptional activation of the SULT1E1 gene, and chromatin immunoprecipitation (ChIP) assays were employed to examine the interaction of RORα and ERα on this PE motif using phosphor-specific antibodies. It appeared that both RORα and ERα were phosphorylated on the SULT1E1 promoter when it was activated in diabetic women livers. The SULT1E1 and output transporter BCRP/ABCG2 coordinately increased their expression levels in the livers of diabetic women but not men. This coordinated expression will be discussed as the mechanism to eliminate active estrogen and as a defense mechanism against estrogen toxicity in diabetic women.

2. Materials and Methods

2.1. Reagents.

Small interfering RNA (siRNA) targeting ERα or RORα and scrambled control siRNA were from GE Healthcare Dharmacon, Inc. (Lafayette, CO). Monoclonal antibody against ERα (#ab32063) was obtained from ABCAM (Cambridge, MA). Polyclonal antibody against RORα, p-RORα (an antibody specific to Serine 100 phosphorylated RORα), and p-ERα (an antibody specific to Serine 212 phosphorylated ERα), were custom-generated by Genescript Biotech (Piscataway, NJ). Monoclonal antibody against RORα (# PP-H3910–00) was purchased from Perseus Proteomics Inc. (Tokyo, Japan). Normal rabbit IgG were obtained from Santa Cruz Biotechnology, Inc. (sc-2027). PCR primers and custom designed oligos (Table 1), William’s E medium, hepatocyte thawing medium (CHRM®, CM7000), hepatocyte plating pack (CM3000), hepatocyte maintenance pack (CM4000), lipofectamine RNAiMAX reagent, High Capacity Archive kit, Power SYBR Green Master Mix (no. A25741), TaqMan Universal Master Polymerase Chain Reaction (PCR) Mix (no. 4304437), TaqMan PCR probes (Hs99999905_m1 for human GAPDH, Hs01060665_g1 for β-actin, Hs00960938_m1 for SULT1E1, Hs00536545_m1 for RORα, Hs01046816_m1 for ERα, Hs01053790_m1 for ABCG2, Hs00223686_m1 for ABCG5, Hs00223690_m1 for ABCG8, Hs00245360_m1 for SLCO1A2, and Hs00272374_m1 for SLCO1B1), Trizol reagent, and Dulbecco’s Modified Eagle Medium (DMEM) were purchased from Life Technologies Corporation (Grand Island, NY). ATP, [γ−³²P] was obtained from PerkinElmer (Waltham, MA). T4 polynucleotide kinase was purchased from New England Biolabs (Ipswich, MA). ChIP-IT Express was purchased from Active Motif, Inc. (Carlsbad, CA). FUGENE® 6 transfection reagent, TNT coupled reticulocyte lysate systems and dual luciferase assay system were obtained from Promega (Madison, WI). Phosphatase inhibitor cocktail-2 and - 3, formaldehyde and 6-(4-Chlorophenyl)imidazo[2,1-b][1,3]thiazole-5-carbaldehyde O-(3,4-dichlorobenzyl)oxime (CITCO), and 2-mercaptoethanol were purchased from Sigma-Aldrich (St. Louis, MO).

Table 1.

Primers and oligos used in chromatin immunoprecipitation and Electrophoretic mobility shift assays.

Application/motif	Forward Sequence: 5’−3’	Reverse Sequence: 5’−3’
ChIP
RORE	GTGAGTTGTCACTCTTATCCTTAC	CTATGTATTTCACTGTGTCTGCAG
ERE	GTCTGTCTGCATATCTGCCCAC	GGCCCTGAAACCTCATTCTTCTCC
PE	TGAATGAACATGACTCTGTGTGG	ATTTGGGGGCTTGGCTTCAT
EMSA
RORE	agaGAATTGAGGTCAcag	ctgTGACCTCAATTCtct
ERE	gagGGTCAagaTAAACatt	aatGTTTAtctTGACCctc
DR4	tccTGACCTCAAGTGATCTgct	agcAGATCACTTGAGGTCAgga

Abbreviations: ChIP: Chromatin immunoprecipitation; RORE, RORα response element; ERE, ERα response element; PE, Proximal element; EMSA, Electrophoretic mobility shift assays, DR, direct repeat

2.2. Human livers.

Human liver specimens were purchased from Sekisui Xenotech, LLC (Kansas City, KS). The diagnosis and classification of the livers as diabetic or nondiabetic was made by the provider, and age, body mass index (BMI), race and diabetic classifications are presented in Table 2. The liver biopsies were categorized into four groups: nondiabetic women (n = 7), diabetic women (n = 12), nondiabetic men (n = 9), and diabetic men (n = 5). Since obesity, defined as BMI greater than 30 kg/m² [29], is a leading risk factor for T2D [30, 31], nondiabetic livers (n = 4) with BMI > 30 kg/m² were included in the diabetic group. Mean BMI in the women and men diabetic groups were 36.5 (standard deviation SD, 5.6) and 39.3 (SD, 5.2), respectively. The corresponding BMI values were 24.2 (SD, 3.3) and 23.1 (SD, 5.2) in the nondiabetic women and men groups respectively (Table 2).

Table 2.

Liver biopsy donor summary – body mass index, age, race, and diabetic status.

	WOMEN			Men
	Nondiabetic	Diabetic	P value	Nondiabetic	Diabetic	P value
BMI, mean (SD)	24.2 (3.0)	36.5 (5.6)	<0.0001	23.1 (5.2)	39.3 (5.2)	0.00012
Age, mean (SD)	56.0 (13.4)	52.8 (8.9)	0.54	49.2 (15.2)	51 (14.2)	0.83
Race
AA	2	6		5	0
CA	4	6		4	3
HS	1	0		0	2
Diabetic	0	8		0	5
Nondiabetic	7	4		9	0

Abbreviations: BMI, body mass index; AA, African American; CA; Caucasian; HS, Hispanic; SD, standard deviation.

2.3. HPH culture and Treatment.

Cryopreserved HPH were recovered per the provider’s guidelines. Briefly, the cryopreserved cells were thawed in a water bath at 37 °C and immediately transferred into hepatocyte thawing medium, centrifuged at 100 x g for 10 min, and the cells were resuspended in hepatocyte plating medium (William’s E medium supplanted with 5% fetal bovine serum, 1 µM dexamethasone, penicillin/streptomycin, human recombinant insulin, GlutaMAX™, and HEPES pH, 7.4), seeded in 12-wells plate at a density of 400,000 cells/well, and cultured for 4 hrs. at 37 °C supplemented with 5% CO₂. The medium was replaced by hepatocyte maintenance medium (William’s E medium augmented with 0.1 µM dexamethasone, penicillin/streptomycin, human recombinant insulin, human transferrin, selenous acid, bovine serum albumin, linoleic acid, GlutaMAX™, and HEPES, pH 7.4), and incubated overnight at 37 °C supplemented with 5% CO₂. siRNA transfection and drug treatment. HPH cells were transfected with control siRNAs, or siRNAs targeting ERα or RORα for 72 hrs. using lipofectamine RNAiMAX reagent per manufacturer’s guide. Concisely, a mixture of siRNA and lipofectamine RNAiMAX reagent was prepared in Opti-MEM® medium and incubated at room temperature for 5 min. 100 µL mixture (final 10 pmol siRNA and 3 μL transfection reagent) was transferred into each control or treatment well, and the transfection was repeated every 24 hrs. for three days. Drug treatment. For the drug treatment, HPH were maintained in hepatocyte maintenance medium and treated with phosphate-buffered saline (PBS), 1mM PB, DMSO, or 250 nM CITCO for 24 hrs. The cells were harvested in Trizol ® Reagent for total RNA extraction or in 1% formaldehyde (FA) for chromatin preparation for ChIP assay.

2.4. RT-PCR.

Total RNA was extracted in Trizol reagents per the manufacturer’s guide from 20 mg frozen liver specimen or cultured HPH, and reverse transcribed by using High Capacity Archive kit (Applied Biosystems) to prepare cDNA. mRNA was quantified using TaqMan universal PCR master mix and probes for SULT1E1, RORα, ERα, ABCG2, ABCG5, ABCG8, SLCO1A2 or SLCO1B1 in CFX96™ Real-time PCR system (Bio-Rad laboratories, Hercules, CA). GAPDH and β-ACTIN were used as house-keeping genes for the expression of target genes in human liver specimen and HPH culture, respectively.

2.5. ChIP assays.

ChIP assays were performed in human liver specimen or HPH using active motif’s ChIP IT EXPRESS kit per the manufacturer’s instructions. Briefly, chopped 100 mg frozen human liver specimen or HPH were cross linked by 1% FA in PBS buffer by incubating at RT for 10 min followed by a 5 min treatment with glycine to stop the crosslinking reaction. The resulting homogenate was incubated in lysis buffer at 4 °C for 30 min. The lysate was centrifuged at 2400 x g, and the nuclear pellet was resuspended in a shearing buffer and the DNA was sheared. 10 µg fragmented chromatin from human liver or HPH was incubated overnight with anti-RORα (# PP-H3910–00), p- RORα, ERα (#ab32063), p- ERα antibodies, or normal IgG in a reaction mixture consisting of G protein–conjugated magnetic beads and protease inhibitor cocktail in a ChIP buffer. The beads were washed, cross-linking was reversed, and the purified DNA were used for PCR amplifications using custom prepared primers (Table 1) flanking RORE, ERE, and PE on the SULT1E1 promoter.

2.6. Electrophoretic mobility shift assays (EMSA).

The response elements for CAR (DR4), RORα (RORE), and ERα (ERE) on the SULT1E1 promoter were tentatively identified using NUBIScan, an In Silico prediction tool, [32], and the interactions between the nuclear receptors and their candidate response elements were studied by EMSA as EMSA were described in our previous studies [26, 33]. Briefly, double-stranded oligonucleotides consisting of DR4, RORE, or ERE (Table 1) were radio-labeled with [γ−³²P]ATP in the presence of T4 polynucleotide kinase. CAR, RXR, RORα and ERα were in vitro translated from mammalian expression plasmids in pcDNA 3.1 using TNT Coupled Reticulocyte Lysate System (Promega) per manufacturer’s guides. The radiolabeled oligonucleotides and the in vitro translated proteins (1µL) were mixed and incubated at room temperature for 10 min followed by electrophoresis on acrylamide gel.

2.7. Reporter assay.

Cos-1 cells were seeded in a 24-wells plate in DMEM medium for 24 hrs. RORα, RORα S100A, RORα S100D, ERα, ERα S212A or ERα S212D mammalian expression plasmids were prepared in pcDNA 3.1 vector. Three reporter constructs namely: construct I, II and III prepared by placing −5.3 kb, −5.3 kb/Δ2, or −177 bp SULT1E1 promoter in front of luciferase gene in a PGL3/basic vector. The −5.3 kb/Δ2 reporter construct was prepared by deleting a proximal promoter element (PE, −120/−54) within the −5.3 kb SULT1E1 promoter. Cos-1 cells were cotransfected with RORαs, ERαs, or their combinations in the presence of reporter constructs and Renilla luciferase. The activity of each promoter in response to RORαs, ERαs or their combinations was determined from relative luciferase activity by the Dual Luciferase Assay System (Promega).

2.8. Statistical Analysis.

All statistical analyses were performed on GraphPad Prism software (version 8.2.1). P values for the observed differences were obtained by using t test (nonparametric tests, Mann-Whitney test), or Multiple t test (Holm-Sidak method, alpha = 0.05). Error bars in all Figures represent mean ± SD (standard deviation).

3. Results

3.1. Hepatic SULT1E1 expression in diabetic humans

Human liver specimen from commercial source were categorized as: nondiabetic (n = 7) and diabetic (n = 12) women livers, and nondiabetic (n = 9) and diabetic (n = 5) men livers. The SULT1E1 mRNA expression was determined in qPCR, and the relative average expression levels of SULT1E1 mRNA (corrected to GAPDH) was 0.00127 and 0.00475 in the liver of nondiabetic and diabetic women, respectively (Fig. 1A) whereas the corresponding levels were 0.00166 and 0.00040 in nondiabetic and diabetic men (Fig. 1B). These results indicated that the expression of the SULT1E1 mRNA in diabetic women was significantly increased by 3.8-fold (p = 0.0052) over the levels observed in the livers of nondiabetic women (Fig. 1A). On the other hand, although the diabetic conditions appeared to decrease the already lower level of the SULT1E1 in the men liver biopsies by 1.9-fold (p = 0.6064) compared to the nondiabetic controls, the change was statistically insignificant (Fig. 1B). There was difference in the relative average expression level of the SULT1E1 mRNA in the women and men livers (Fig. 1C) which appeared to be driven by an elevated expression of the SULT1E1 gene in diabetic women compared to the diabetic men livers (Fig. 1D). The SULT1E1 expressions were analyzed based on age. As a probable reflection of the different incidences in the development of T2D between women and men, liver samples used accumulated around 50–60-year-old diabetic women, whereas those of diabetic men were scattered through their ages (Fig. 1E). Therefore, the increase in the hepatic SULT1E1 expression appeared as a characteristic in diabetic women around menopause.

Figure 1. Expression of the SULT1E1 mRNA in human liver.

Total RNA was extracted from human liver specimen and reverse transcribed to cDNAs using High Capacity Archive kit. The SULT1E1 mRNA relative expression (corrected to GAPDH) was quantified by using qPCR. A. Expression of the SULT1E1 in non-type 2 diabetic (N-T2D) controls (n = 7) and T2D women livers (n = 12). B. Expression of the SULT1E1 in N-T2D (n = 9) and T2D (n = 5) men livers. C. Comparison of relative SULT1E1 mRNA expression in women (n = 19) and men (n = 14). D. SULT1E1 mRNA expression in T2D women (n = 12) and T2D men (n = 5). E. The relative SULT1E1 values were plotted against the age of the liver donor. Symbols: , women; , men; , nondiabetic women; , diabetic women; , nondiabetic men; , diabetic men. F. Total RNA was extracted from human primary hepatocytes (HPH) treated with phosphate-buffered saline (PBS) or phenobarbital (PB) for 24 hrs. and the SULT1E1 mRNA was quantified by using qPCR. The asterisk (*) represents statistically significant difference (*p < 0.05; **p < 0.001; ***p < 0.001) in the expression the SULT1E1 mRNA between nondiabetic and diabetic groups or PBS and PB treated HPH.

Having these findings, HPH were utilized to investigate the regulatory mechanism of this SULT1E1 expression. It was found that HPH express the SULT1E1 mRNA as found in diabetic women (Fig. 1F) and PB treatment repressed this SULT1E1 expression in HPH like what was observed in nondiabetic women. Consistent with these observations, the expression of Sult1e1 mRNA was significantly increased in the diabetic mouse model Ob/Ob livers although no gender-specific differences were observed in this mouse model (Data not shown).

We also examined whether diabetic conditions alter the expressions of selected members of ABC (ABCG2, ABCG5, and ABCG8) and SLCO (SLCO1A2 and SLCO1B1) membrane transporters. Diabetic women, but not men, livers upregulated the expression of ABCG2 mRNA compared to non-diabetic controls (Fig. 2A and B). Moreover, the expression ABCG2 mRNA correlated with the expressions of STUL1E1 mRNA in diabetic women’s liver biopsies (Fig. 2C). The diabetic conditions did not change the expressions of SLCOB1 mRNA both in the women and men liver biopsies, but its expression was correlated with that of SULT1E1 in the women livers (Fig. 2D, E, & F). Like ABCG2, the diabetic livers increased the expression of ABCG5 in the women liver biopsies (Fig. 2G). We did not observe statistically significant difference in the expressions of ABCG8 and SLCO1A2 between diabetic and nondiabetic livers both in the women and men group (Fig. 2H-L).

Figure 2. Expression membrane transporters in diabetic and nondiabetic livers.

Total RNA was extracted from liver biopsies in Trizol reagent and cDNA was synthesized from the resulting RNA. A & B. Relative expression of ABCG2 in women and men liver biopsies. C. Correlation between ABCG2 and SULT1E1 mRNA expression in women liver biopsies. D & E. Relative expression of SLCO1B1 mRNA in women and men livers. F. Correlation between SLCO1B1 and SULT1E1 mRNA expression in women liver biopsies. G, H, & I. Relative expression of ABCG5, ABCG8, and SLCO1A2 in nondiabetic and diabetic women livers. J, K, & L. Relative expression of ABCG5, ABCG8, and SLCO1A2 in nondiabetic (N-T2D) and diabetic (T2D) men livers. The asterisk (*) represents statistically significant difference between groups (*p < 0.05; **p < 0.01).

3.2. SULT1E1 Regulation by RORα and ERα in HPH:

PB induced the Sult1e1 mRNA expressions in the mouse liver and this induction is co-regulated by nuclear receptors Car, Rorα, and Erα [25, 26]. Conversely, the SULT1E1 mRNA was found to be constitutively expressed in HPH and was repressed following PB treatment (Fig. 1F). Hypothesizing that these nuclear receptors may also regulate the SULT1E1 expression and/or repression in HPH, siRNAs were employed to examine whether RORα and ERα regulated the human SULT1E1 gene in HPH. Downregulation of RORα mRNA by siRNA diminished levels of the SULT1E1 mRNA in HPH (Fig. 3A&B). Similarly, siRNA against ERα also decreased the SULT1E1 mRNA expressions in HPH (Fig. 3C&D). Since siRNA did not work to downregulate CAR, we used a reverse CAR agonist CITCO which demonstrated that CAR repressed the SULT1E1 mRNA expression in HPH (Fig 3E). Besides, overexpression of CAR in Cos-1 cells repressed a −5.3 kb SULT1E1 promoter in a reporter assay (Fig. 3F). These findings indicated that RORα and ERα activated the SULT1E1 gene while CAR repressed it in HPH.

Figure 3. RORα and ERα regulate the SULT1E1 gene in human primary hepatocytes (HPH).

HPH cultures were treated with control siRNAs or siRNAs targeting RORα or ERα for 72 hrs. At the end of treatment, the cells were harvested, total RNAs was extracted, and cDNAs prepared using High Capacity Archive kit. mRNA was quantified in RT-PCR. Relative expression of RORα (A) and SULT1E1 (B) mRNA in HPH cells treated with control siRNA or RORα siRNA. ERα (C) and SULT1E1 (D) mRNA in cells exposed to control or ERα siRNA. E. Total RNA was extracted from HPH cultures treated with DMSO or 250 nM CITCO for 24 hrs., and mRNA was quantified by using RT-PCR. F. The SULT1E1 reporter was transfected into Cos-1 cells in the presence of mock (pcDNA3) or expression plasmid for nuclear receptor CAR and RXR. Dual reporter assay was performed per manufacturer’s guide (Promega). Statistically significant difference between control and siRNA or treatment groups is represented by asterisk (*p < 0.05; **p < 0.01; ***p < 0.001; **** < 0.0001).

3.3. CAR, RORα, and ERα binding motifs within the SULT1E1 promoter.

A motif search was conducted within the SULT1E1 promoter against a potential CAR, RORα and ERα binding elements. CAR binds a DR4 element on its target genes [34], and RORα binds a DNA sequence 5’-AGGTCA-3’ preceded by an A/T rich sequence [35] while ERα binds an inverted repeat of 5’-GGTCA-3’ separated by three nucleotides [36]. This search identified putative DR4, RORE, and ERE at −42987/−42971, −3816/−3805 and −1868/−1856 bp, respectively (Fig. 4A). To confirm the direct binding of CAR, RORα, or ERα to their respective putative motifs, EMSA was performed using in vitro translated CAR, RORα or ERα proteins and double-stranded radiolabeled DR4, RORE or ERE oligonucleotides, respectively (Fig. 4B). The three proteins shifted nucleotides, and while the anti-RORα antibody blocked this shift, anti-ERα antibody super shifted the ERα-nucleotide complex (Fig. 4B). Subsequently, a reporter construct was prepared by placing a 3x-RORE sequence in front of a luciferase reporter gene in a PGL3/basic vector, and a cell-based reporter assay in Cos-1 cells confirmed that RORα activated the RORE (Fig. 4C). Similarly, ERα activated a reporter construct carrying 3x-ERE in PGL-3/basic vector in Cos-1 cells (Fig. 4D). Supporting these observations that RORα and ERα regulate the SULT1E1 in HPH as described (section 3.2), it appeared that RORE and ERE were functional elements regulated by RORα and ERα, respectively. The functionalities of RORE and ERE were examined within the context of the −5.3 kb SULT1E1 promoter (named promoter I) placed in PGL-3 vector. This −5.3 kb promoter comprises of RORE, ERE, PE (−120/−54 bp), and TATA box. Promoters II and III were constructed by internally deleting the proximal element PE, and the whole upstream region of PE (−5.3 kb/−178 bp), respectively (Figs. 5A and 5B) within the −5.3 kb promoter. Promoter II lost its activity while promoter III retained only 20% of the activity observed with the promoter I. These results indicated that RORE and ERE were functional within the context of the SULT1E1 promoter and required the PE to confer their activities. On the other hand, although the newly identified CAR binding motif was located upstream of the −5.3 kb promoter, CAR-RXR complex could still suppress the −5.3 kb promoter (Fig. 3F).

Figure 4. Identification and characterization of response elements for RORα (RORE) and ERα (ERE) within the SULT1E1 promoter.

A. New putative response elements for CAR (DR4, −42987/−42971, RORα (RORE, −3816/−3805) and ERα (ERE, −1868/−1856) within the SULT1E1 promoter. B. Double-stranded oligonucleotides consisting of DR4, RORE, or ERE were end-labeled with [γ−³²P]ATP in a T4 polynucleotide kinase catalyzed reaction. The radio-labeled nucleotides consisting of DR4, RORE, or ERE motifs were incubated with in vitro translated CAR-RXR, RORα or ERα proteins, respectively. Mock (pcDNA3.1) was included as a negative control for the proteins, and consensus CAR, RORα or ERα binding sequences were included as a positive control for the DNA-protein interaction. Anti-RORα (aRORα) and anti-ERα (aERα) antibodies used to confirm the specificity of the band shift observed. C. 3x-RORE sequence was placed in front of PGL3 vector and cotransfected with mock or RORα expression plasmid into Cos-1 cells. Renilla luciferase plasmids were included as an internal control. D. A reporter construct carrying ERE (3x) sequence in front of luciferase gene in PGL-3 vector and mock or ERα expression plasmids were cotransfected into Cos-1 cells in a reporter assay. The asterisk (*) represents statistically significant difference (****p < 0.0001) between mock and RORα or ERα transfected groups.

Figure 5. RORα and ERα regulate the SULT1E1 promoter in Human primary hepatocytes (HPH).

A. Promoter analysis was performed to determine the critical motifs within the −5.3 kb SULT1E1 promoter, and accordingly three reporter constructs carrying: (I)-5.3 kb SULT1E1 promoter; (II)-5.3 kb SULT1E1 promoter with internal deletion of PE motif; and (III) −177 bp SULT1E1 promoter were prepared in PGL3/basic vector. B. Reporter assays were performed in Cos-1 cells and the activity of each promoter was compared against promoter I. C&D. Chromatin immunoprecipitation (ChIP) assays in human primary hepatocytes (HPH). HPH cells were treated with phosphate-buffered saline (PBS) or phenobarbital (PB) for 24 hrs., and ChIP-DNAs were prepared using ChIP-IT Express kit (Active motif). Immunoprecipitations were carried out with two sets of antibodies – general anti- RORα and -ERα antibodies and custom-made antibodies specific to Ser100 phosphorylated RORα or Ser212 phosphorylated ERα antibodies. Immunoprecipitated DNA was amplified with primers targeting RORE, ERE, or PPE motifs. Statistically significant difference is represented by Asterisk (****p < 0.0001) between groups.

Having characterized the interaction between the three nuclear receptors and their newly identified response elements, we utilized ChIP assays to examine the binding of RORα and ERα to these motifs (i.e. RORE, ERE, or PE) in HPH treated with PBS or 1 mM PB. PB repressed the SULT1E1 mRNA expression in HPH (Fig. 1F). Using anti-RORα and custom-made p-RORα antibodies, it was found that RORα bound RORE as well as PE motifs, and the receptor was phosphorylated at Ser100 in untreated cells. In these cells, the SULT1E1 gene was active. However, RORα was released from RORE and PE motifs in cells treated with PB (Fig. 5C), and the SULT1E1 promoter was inactive. Similar ChIP assays with anti-ERα and p-ERα antibodies revealed that ERα was always retained on ERE and PE motifs (Fig. 5D) irrespective of PB treatment. However, the receptor was Ser212 phosphorylated following PB treatment. This was consistent with a previous observation that PB induced Ser212 phosphorylation on the Sult1e1 promoter in mouse liver [25].

3.4. The SULT1E1 promoter activation in diabetic women.

Having demonstrated that the phosphorylation status of Ser100 RORα and Ser212 ERα residues dictated the SULT1E1 mRNA expression in human HPH, the proximal element PE was critical in the transcriptional regulation of SULT1E1 by nuclear receptors ERα and RORα, and the fact that the diabetic conditions induced the hepatic SULT1E1 expression in women (Fig. 1A), we utilized ChIP assays to examine the interaction of these receptors with the PE motif. Towards this end, chromatins were prepared from the livers of four nondiabetic and four diabetic women. Whereas anti-RORα antibody revealed that the receptor occupied the PE motif in both nondiabetic and diabetic livers, the receptor was increasingly phosphorylated at Ser100 in the diabetic livers compared to nondiabetic livers per ChIP data with anti- p-RORα antibody (Fig. 6A). We also quantified the RORα binding to the PE motif by qPCR assays. Levels of RORα on the PE motif did not show a significant change between diabetic and nondiabetic livers (Fig. 6B). However, p- RORα antibody data revealed that Ser100 phosphorylated RORα enrichment on this motif was significantly increased (Fig. 6C). ChIP data with anti-ERα antibody showed ERα binding on the PE motif both in the nondiabetic and diabetic livers (Fig. 6D). ERα detected by p- ERα antibody revealed an increase in the Ser212 phosphorylated ERα on the PE motif in diabetic over nondiabetic women livers (Fig. 6D). Subsequent qPCR confirmed this increase in diabetic livers (Fig. 5E and 5F). These results indicated that both RORα and ERα were phosphorylated on the PE motif when the SULT1E1 promoter was active in the livers of diabetic women, and the level of the SULT1E1 mRNA positively correlated with the Ser100 and Ser212 phosphorylated RORα and ERα, respectively, enriched on the PE (Fig. 6G and 5H). Supporting this observation, phosphomimetically mutated ERα S212D and RORα S100D activated the −5.3 kb SULT1E1 promoter in a cell-based reporter assay. Neither ERα S212D nor RORα S100D mutant expression plasmids were active against the SULT1E1 promoter in Cos-1 cells (Fig. 6I).

Figure 6. Ser100 phosphorylated RORα and Ser212 phosphorylated ERα enrichment on the SULT1E1 promoter.

Chromatin immunoprecipitation (ChIP) assays were performed in nondiabetic (N-T2D, n=4) or diabetic (T2D, n=4) women livers and immunoprecipitations were carried out with anti- RORα and -ERα antibodies and custom-made antibodies specific to Ser100 phosphorylated RORα (p-RORα) or Ser212 phosphorylated ERα (p-ERα) antibodies. The immunoprecipitated DNA was amplified by primer sets targeting the proximal promoter region (PE). A. ChIP-DNA immunoprecipitated by anti- RORα or p-Ser100 RORα antibodies were amplified and electrophoresed in a 2% agarose gel. B&C. The relative enrichments of DNA fragments immunoprecipitated by anti- RORα (B) and p- RORα (C) antibodies on the PE motif were quantified in RT-PCR using PCR primers targeting the proximal motif. D. Chromatins immunoprecipitated by anti- ERα or p- ERα antibodies were amplified and electrophoresed in a 2% agarose gel. E&F. Fold enrichments (relative to IgG) of the ERα immunoprecipitated by anti- ERα (E) or anti- p- ERα (F) on the PE motif were quantified in qPCR assay. G&H. Correlation between the expression level of the SULT1E1 mRNAs and enrichment of p- RORα (G), and p- ERα (H) on the PE motif on the SULT1E1 promoter. I. Reporter assay in Cos-1 cells. Cos-1 cells were transfected with phosphomimetically mutated ERα S212D, RORα S100D, or both in the presence −5.3 kb SULT1E1 promoter. Asterisk represent statistically significant difference (**p < 0.01) between the diabetic and nondiabetic groups.

4. Discussion

T2D women upregulate SULT1E1expression in the liver thereby increasing the metabolic capability to inactivate estrogens via sulfation. This upregulation was specific to women and was not observed in diabetic men. In response to diabetes, the SULT1E1 gene is activated in diabetic women livers. Nuclear receptors ERα and RORα are sequentially phosphorylated at their conserved motifs, and then interact with, and activate the SULT1E1 promoter. Diabetic livers also upregulated the expression of ABCG2 which exports sulfated estrogens into the bile duct, tightly linking inactivation to the export of estrogens.

The SULT1E1 gene contains ERE and RORE motifs in a distal region of the promoter. When the gene is repressed, ERα and RORα bind ERE and RORE, respectively, of which RORα is the one that represses SULT1E1 promoter activity, while ERα appears to be non-functional. RORα’s functionality as a transcriptional repressor is consistent with previously reported findings with Sult1e1 and Cyp2b10 genes in mouse livers [26, 37] and in HepG2 cells cultured in high glucose (450 mg/dL) medium [38]. ERα and RORα conserve a phosphorylation motif at Ser212 and Ser100, respectively, within their DBD [39, 40]. On the other hand, when the gene is activated in diabetic livers, these motifs were occupied by Ser212 ERα and Ser100 RORα phosphorylated receptors, and interacted with the proximal element PE, possibly as a complex. Moreover, the degrees of phosphorylation of Ser212 ERα and Ser100 RORα on the PE motif is well correlated with expression levels of SULT1E1 mRNA in women livers. This activation mechanism of the SULT1E1 gene by phosphorylated ERα and RORα resembles that of the Sult1e1 gene in diabetic mouse livers [25, 26]. In Sult1e1 gene activation, RORα was not phosphorylated in the liver of ERα S216A KI mice, indicating that ERα phosphorylation is a prerequisite to the subsequent phosphorylation of RORα. Thus, ERα initiates this sequential phosphorylation to enable phosphorylated RORα to activate the Sult1e1 promoter. Therefore, this ERα-initiated mechanism could be conserved in the livers of diabetic women, and a diabetes response signal must be transduced to ERα as it initiates activation of the SULT1E1 gene.

Although it is difficult to directly confirm signals in human livers, what was found with mouse livers should provide an insight as to how the SULT1E1 activation is signaled. Acting as a xenobiotic signal, PB induced the activation of the Sult1e1 gene in mouse livers [25, 26]. PB signal is transduced to nuclear receptor CAR that recruits ERα to the promoter for subsequent phosphorylation. Notably, the ERα interacted with CAR was an estradiol bound form. Therefore, PB integrates its xenobiotic signal with an endogenous estrogen signal to activate the Sult1e1 gene by the function of CAR and ERα. In diabetic mouse livers, since CAR is spontaneously activated, estrogen may be a primary signal to activate the Sult1e1 gene [25]. Confirming these findings, ovariectomy repressed Sult1e1 expression in diabetic mouse livers (unpublished data). These observations implicate estrogens as the primary signal to ERα in diabetic livers of post-menopausal women. As a result, the ERα-RORα-SULT1E1 pathway may constitute a feedback mechanism against estrogens in diabetic women. At present, there are no experimental methods to confirm whether CAR is involved in the activation of the SULT1E1 gene in the livers of diabetic women. However, the fact that CAR activation repressed the gene in human primary hepatocytes suggests the possibility that CAR is also regulating it in diabetic women livers. If CAR represses the SULT1E1 gene, therapies with CAR-activating drugs such as PB may decrease SULT1E1 expression levels causing negative consequences in diabetic women.

In addition to the SULT1E1 enzyme, hepatic membrane transporters also play a role in the metabolic homeostasis of estrogens and their conjugates [22, 41–43]. OATP1B1/SLCO1B1 imports sulfated estrogens from the bloodstream while BCRP/ABCG2 exports into the bile duct [22, 42, 43]. ABCG2 mRNA exhibited a clear increase and had a tighter correlation with the expression of SULT1E1 in women livers suggesting that the ABCG2 gene could be under ERα-RORα regulation as observed with the SULT1E1 gene. ERα is known to positively regulate the ABCG 2 gene in cancer cells [44, 45]. Thus, the diabetic liver induces SULT1E1 and ABCG2 to link metabolic inactivation and export of estrogens, thereby increasing its capability to eliminate estrogens.

The association between the level of circulating estrogens and T2D in postmenopausal women remains controversial. Although the beneficial role of estrogens in energy metabolism is well established in premenopausal women [7, 8, 46, 47], several epidemiological studies suggested that higher circulating estrogens are biomarkers of T2D risk in postmenopausal women [48–50]. Thus, increased SULT1E1 and ABCG2 expression in the livers of diabetic women can be of pathophysiological significance. Estrogen-ERα-RORα-SULT1E1 constitutes a defense signal that inactivates estrogens, and the increased expression of ABCG2 completes the metabolic elimination of sulfated estrogens. In mouse T2D model, expression of Sult1e1 was reportedly detrimental as ablation of the Sult1e1 gene improved metabolic conditions including energy expenditure, insulin sensitivity, and inflammatory complications in female Ob/Ob mice, an animal model for type 2 diabetes mellitus while ablation of the Sult1e1 gene aggravated the diabetic conditions in male Ob/Ob mice [24, 51]. This suggested the Sult1e1 may have a gender-specific functions in energy metabolism. Here, in the livers of diabetic women, the increased SULT1E1 expression appeared to be beneficial for diabetic women. Similarly, increased expression of SULT1E1 in ovary and breast are reportedly beneficial against estrogen induced oncogenesis [12, 52, 53]. This enigma may arise from the fact that, unlike postmenopausal women, female mice maintain circulating estrogens. Therefore, what was observed with Ob/Ob female mice has no relevance to type 2 diabetic women. However, despite these phenotypical differences, the nuclear receptor-mediated mechanism that regulates the SULT1E1 gene is conserved in both mice and humans [25, 26].

In conclusion, T2D women coordinately increased hepatic expressions of SULT1E1 and ABCG2 thereby linking estrogen metabolism and export to eliminate estrogens. The SULT1E1 gene was activated in response to diabetes in women livers. ERα may sense excess estrogen levels to promote RORα to activate the SULT1E1 promoter. This ERα-RORα regulation may also be conserved to activate the ABCG2 gene. Thus, the liver is equipped with a defense mechanism against estrogen toxicity in diabetic women.

5. Abbreviations

AA

African American

ABCG2

ATP Binding Cassette Subfamily G Member 2

ABCG5

ATP Binding Cassette Subfamily G Member 5

ABCG8

ATP Binding Cassette Subfamily G Member 8

CA

Causasian

CAR

Constitutive androstane receptor

ChIP

Chromatin immunoprecipitation

EMSA

Electrophoretic mobility shift assay

ERE

Estrogen receptor response element

ERα

Estrogen receptor alpha

HS

Hispanic

IR3

Inverted repeat of 5’-GGTCA-3’ separated by three nucleotides

p-

Phosphorylated

PB

Phenobarbital

PBS

Phosphate-Buffered Saline

PE

Proximal element

RORE

Retinoid-related orphan receptor alpha response element

RORα

Retinoid-related orphan receptor alpha

siRNA

small interfering RNA

SLCO1A2

Solute Carrier Organic Anion Transporter Family Member 1A2

SLCO1B1

Solute Carrier Organic Anion Transporter Family Member 1B1

SULT1E1

Estrogen sulfotransferase